Smart Lending — Modernizing Legacy Systems through API-led Scalable Architecture

7 min readApr 9, 2021

Background

We are living in an unprecedented era of new challenges, COVID revamps multiple fronts of the financial industry from SME to large scale business to individuals. But In response to the global COVID-19 recession, central banks across the world unleashed synchronized monetary stimulus to backstop the economy, driving short-term and long-term rates downward. This has brought the opportunity for long term borrowers to collect real states. Global interest rates will likely stay hang on with current low levels for at least the next couple of years, perhaps maybe more because central banks globally will feel little pressure to raise interest rates any time soon.

Short-term interest rates are determined by central banks. Longer-term interest rates, like the U.S. 10-year Treasury yield, are determined by the average expected short-term rate over the next 10 years, plus some term premium. Thus, central bank actions, and guidance for future actions, will directly impact short-term and long-term yields.

What it means for the next few years;

Average inflation will delay rate hikes:
No inflation pressure during early cycle recovery:

With this movement, we have seen a shift in the financial industry, especially lending under financial sector. As someone who lives in the Bay area (owning a home) ~ like me, COVID opened flood gates for many opportunity seekers, I’m getting daily promotion mails/calls from various lenders to consider refinancing because of the historically low-interest rates, please refer following charts. (The lowest ever in the history) http://www.freddiemac.com/pmms/

When looking this problem from a business perspective, this opens huge opportunities for lenders as well as borrowers to compete and onboarding customers by offering competitive rates with attractive 15 years, 30-year terms.

What are the challenges Lenders facing?

Based on various states collected through multiple sources, I learned that lenders face immense pressure from surging mortgage demand, with government backup forgiveness programs and massive small business relief programs.

When someone considers a mortgage, the process starts with the borrowers; these are real people who would like to buy their first home or try to refinance their mortgage or investors trying to increase their inventory. However, once you get onboard sail through the process is a real challenge because, borrower and leader has to deal with multiple 3rd party agencies such as, credit buro’s, title company, to real estate agents, property appraisal agencies, electronic Mortgage Affiliates as Ellie mae, employment verification (VOE) during underwriting review, which is time-consuming.

For traditional lenders, with borrower demand and the competition, the borrower allows to shop around multiple lenders, whoever offering the best rate, attractive package without costing too much points to close the deal within a short period wins the race.

Here the problem is, lenders who have a traditional mindset with strict regularity focusing business would face huge setback due to lack of adaptability because of the monolithic traditional business models. Due to that reason, the competition and time become the most critical value to measure the outcome

There are lenders such as better.com, Lendingtree, loanDeport (not intended sponsoring) but they all have state of art digital landing experience through seamless onboarding to underwriting to seal the deal (How I know this, because, I went through some of them, at least through the evaluation phase).

Normally, the average closing of the lenders in 2020 stretches around 46 days within the United States per 2020 data (which is still too high), but I managed to seal the deal with one of the lender who has provided a complete digital experience and semless smooth underwriting to funding within 14 days, the experience I called digital transformation.

What it means, if any borrower start the process with the traditional lenders (Like traditional banks), they are stuck in between the integration points (I had this experience 2 years ago), do all the manual work for the lender, which means frustration, that’s why millennials are looking to FInTechs to none financial organization to solve this problem, that’s where we have seen the blossom of online lenders such as better.com, loanDeport etc etc

Why Data process silos hurt team productivity for traditional lenders

A recently completed survey revealed 73% of the lenders think that the sales process of uncovering customer needs is inefficient simply because, the borrower information lives through multiple subsystems and disconnected systems, thus it requires manual intervention. There are whole sort of problems upfront to deal with

The Delivery capacity- CIO responsibility
Borrower 360 references which include service orchestration
Integrate with multiple CRM systems for customer portfolio validation
Credit buro integration
E-signature verification for underwriting -COVID impact
eClosing experience with employment verification with COVID bubble such as integrate with tools like DocuSign.
Mobile access, one-click process (mostly)

Let’s look at typical lender’s monolithic architecture;

If you look at systems architecture above, let assume it developed decades ago, which means it adds lots of complexity when adjusting to modern technologies due to none conventional approach. Some of the common issues they would see as adding new products, data stores, such as changing pricing strategies, introduce modules to adapt to the technology revamp.

With well defined layered architecture, it allows someone to deploy fully distributed deployment solution to adhere to the future growth and the expansion.

Modernizing legacy systems

Let’s look at how digitalization would help to overcome this legacy monolothic problem;

Followings are the main strategies we should identify;

Decomposing the monolithic systems
Implementing reusable, solutions
Shifting from close model to the open model through ecosystem modularization

The decoupling of business process, the user experience, and the system integration with three-layered architecture called API centric integration this categorizes into, System, Processes, Experiences

Considering the above diagram, let’s look at how to normalize the integration architecture.

We can start with a System’s of records to abstract the complexity with system APIs, which help the process layers to consume those System API data. The Process APIs can individually execute loan origination (mortgage process) inter connected with underwriting APIs, Fiance APIs, To calculate Interest rates, while communicating through recommendation services after having reviewing borrower credit worthiness. These all can be microservices with inter-services interaction with modularization. Then the Experienced APIs provide a set of APIs to end users to access from WebAPI, from Agent APIs, from Mobile API

Solution Architecture

The experience API layer can be matched with managing the APIs with the API Management product. The APIM offers capabilities such as security, monitoring, rate-limiting, throttling, caching, monitoring. When mapping APIM requirements to the lender capability, the agent APIs, Web APIs, and Mobile API accessibility can be managed centrally with the aforementioned API capabilities while incorporating developer, consumer experience. Also, the Gateway architecture allows the scalability and multiple deployments such as on-premises, cloud, hybrid through CICD practices targetting each LOB (Line Of Business) capabilities.
The process layer integrates with stateful and stateless services. As depicted in the above diagram, process APIs depending on the service orchestration between inter-services communication to fulfill the given process. The process layer not necessarily requires to pack with API-led integration, but still, organizations can have the ability to integrate APIM if they wish to control the inter-process communication. Which can be easily built with the integration. Process APIs can use interprocess communication protocols such as gRPC, HTTP, or can integrate with Kafka, NATS for asynchronous communication. The integration process can be run through the recommendation services while analyzing borrower demand after fullfill client credit worthiness while communicating with the System APIs
System API helps the process APIs to communicate with the external services to full fill the business operations, such as, using connector APIs, involve 3rd party systems invoke requested result to validate. System APIs layer can be developed as microservices which is more aligned with the container-native deployment model and technology like spring-boot can be used for this layer. WSO2 micro integrator can also be considered at this layer if the amount of business logic needs to implemented at this layer is considerably low.

How Open source technology can help to achieve the goal (WSO2)

Open-source technologies have allowed people to build this sort of larger project with small steps without burning a lot of their IT budgets/funds, this gives the total cost of ownership, also due to the nature of open source they can customize and can use to fit the budget. If your enterprise is moving towards cloud-native, container-based infrastructure, the same above-mentioned solution architecture can be accomplished with the WSO2 product suite and can be deployed on-premises or private cloud or hybrid.