Three Myths Why Not to Invest in a Decision Engine
Introduction
In a two part series of articles last month we examined what is a decision engine and what are the core components and functionality? This article will review the top three myths and excuses that credit grantors make to avoid investing in such an indispensable component of their lending toolkit.
Please note that the focus of this article is on a Credit Decision Engine operating in the Originations environment.
The Top Three Myths
Over time a vendor hears many different reasons why a prospect cannot proceed with what is considered a “no brainer” decision. Whilst some of the excuses are just plain daft, we have narrowed down the three most ‘reasonable’ excuses.
The top three excuses as to why not to invest in a decision engine are as follows. We will describe these excuses and then debunk the flawed logic that they contain and expose them for the myths that they are:
- We have a Loan Originations System/Loan Management System
- We have credit scoring
- All of our systems are on-premises only
We have an LOS/LMS already!
Many modern Loan Origination Systems (LOS) include rudimentary functionality for the risk assessment of a credit application. Simple policy rules, scorecard implementation and cut-offs enable the LOS to be a basic end-to-end solution for account origination.
Some LOS salesmen will tend to ‘over-sell’ the true capabilities of their application processing systems and so when they state that they can provide full decisioning, they will be believed by small and entry-level credit grantors.
A Loan Origination System (LOS), also known as an Application Processing System (APS), is a complex software tool that performs a lot of heavy lifting. An LOS manages the credit application from inception to completion, including all of the following tasks:
- Control the process of each application
- Gather all of the data via application screens, internal and third-party files
- Queue applications for manual review, if required
- Provide for document upload, storage, and retrieval
- Feed the final decision back to the applicant
- Liaise with the loan management system to set up approved accounts
Most LOS include the basic components for the basic assessment of the risk of an application. However, given the breadth of functionality that an LOS must provide, the decisioning capability housed within a LOS is entry-level at best, and inadequate for the needs of a modern credit grantor.
Where a decision engine really sets itself apart from an LOS is not in the ability to implement complex strategies (although this is important), but in the ability to help the credit grantor to grow and learn from their credit strategies.
For most lenders, portfolio profitability is directly linked to the credit risk decisions that are made. The ability to track, analyse and improve on credit strategies is key to identifying areas where the strategies are not optimal, and this enables the testing of innovative ideas for improvement.
A good decision engine will provide credit grantors with the ability to simulate new strategies before deployment, and to assign multiple A/B strategies randomly, in parallel. By following this approach, lenders can test major or minor tweaks to their credit granting strategies in order to address targeted profitability levers.
Through well-constructed tests and clean A/B tests deployment, it is certainly possible to simultaneously grow revenues and reduce bad debt, and grow portfolio profitability significantly, year on year. (This is a major benefit of the ‘test and learn’ approach).
Ultimately, the question is a simple one for credit grantors. Are credit granting decisions a key driver in your business’ profitability? If the answer is ‘yes,’ then the business should be using the best available tools to continually improve those decisions.
Only a decision engine provides the required decisioning functionality to understand and optimise a lender’s credit decisions.
We have credit scoring!
Application scoring has been around since the 1960s and by the dawn of the new century most markets in the world had accepted the overwhelming benefits and advantages of using an application score over manual and judgemental decisions.
Unfortunately in a number of frontier markets, unsophisticated lenders mistake the process of calculating application scores and then generating an Accept/Refer/Decline decision as full decisioning. Hence the confusion as to why do they require a dedicated decision engine in addition to application scorecards?
Later, as this century progressed, artificial intelligence matured into machine learning and the role of the data scientist emerged with new and powerful model building and model execution tools readily at hand. Machine learning techniques have enabled the life cycle for model building to become greatly accelerated. In addition, the scope of the actual models has been greatly expanded.
The output of machine learning model building is now often not just a simple formula. Advanced tools allow for more sophisticated feature engineering, enabling the extraction of valuable insights from a wider range of data sources. This can potentially enhance the predictive power of credit risk models by incorporating non-traditional variables.
With the advent of more sophisticated data science tools and techniques, the reality is that a new standard for model implementation has arisen. This involves moving the model execution from within the decision engine to behind a well-managed and scalable endpoint. Each model offers its own endpoint.
When a data scientist sees the computational power behind an endpoint, there is a temptation. Taken to the extreme, they will ask, do we really need a decision engine if we can instead move the decision into the model execution components?
A little thought will demonstrate that this approach has many disadvantages. The decision process is more complex and rightly belongs under the control of the credit risk manager, who is responsible for managing the portfolio(s) and has more knowledge and understanding of credit policy.
The practice of segregating model execution and decisioning, by serving models via service endpoints and employing a configurable decision engine, is a prudent and advantageous strategy for organisations leveraging AI and machine learning. It enhances efficiency, transparency, flexibility, security, and compliance, all of which are essential factors for successful AI implementation.
As we continue to rely on AI and ML models to drive business decisions and optimise processes, embracing this segregation approach is a step towards responsible and effective AI deployment, ensuring that the technology serves the best interests of both credit grantors and society at large.
All of our systems are on-premises only!
The benefits of cloud computing over on-premises are numerous, and to list the top three:
“Cost efficiency. Since the cloud service provider covers all costs connected with the servers’ maintenance and management, their clients may save up the funds they would otherwise spend on on-prem data centres.
Scalability. The companies can scale the cloud resources they use up and down, depending on their needs and business processes.
Mobility. All the computing resources are available on-demand, so the companies can enjoy resiliency and elasticity when accessing them at any time, from anywhere.
Some of the shortcomings of on-premises computing concern the following:
High ownership costs. Reliable on-premises infrastructure requires considerable investments such as updating and maintenance costs, periodic subscription costs, plus the spending on an in-house IT team.
The need to maintain and regularly update the server hardware’s limited scalability. On-premises resources cannot typically be scaled up and down easily—this, too, requires some time and investments.
Hybrid Infrastructure (Cloud And On-Premises)
To enjoy the advantages of both models – cloud and on-premises – some companies opt for the so-called hybrid infrastructure. According to this model, the company integrates a public cloud with its on-prem resources (and sometimes also with a private cloud). So, it has full control over all the data centres and virtual machines. This model, too, comes with its advantages and limitations.
Some of the advantages of a hybrid infrastructure include:
- Scalability
- Adaptability
- Agility
- Elasticity
- Flexibility
Since the hybrid model offers the best of the cloud, all these benefits are included by default. The company can scale computing up and down on a needed basis. From the cost perspective, it can save up a decent amount that would otherwise be spent on maintaining on-prem resources and an in-house IT team.
Hybrid cloud solutions also address the concerns of data sovereignty, which is of prime importance in some countries and is often used as an excuse not to implement a cloud-based decision engine.
In my own personal opinion, I view the reticence by IT managers to embrace cloud computing, even with all of its proven advantages, as similar to fear of flying. Whilst commercial aviation has been around since the 1960s and 70s and has been proven to be far safer and more cost effective than most other modes of transport, there are still people who will not embrace this technology.
Summary
The purpose of this article was to debunk three of the biggest myths around now investing in a credit decision engine.
Please refer to four additional articles, written by three colleagues, which go into more detail as to why the first two excuses and the whole issue around data sovereignty are indeed myths!
Myth 1
Why do lenders require a Decision Engine AND a Loan Origination System?
Myth 2
Separating Predictions from Decisions: Model Execution and Decision Engine
Myth 3
The Data Sovereignty Dilemma: Challenging the Status Quo in Rwanda – Part 1
The Data Sovereignty Dilemma: Challenging the Status Quo in Rwanda – Part 2
About the Author
Stephen John Leonard is the founder of ADEPT Decisions and has held a wide range of roles in the banking and credit risk industry since 1985.
About ADEPT Decisions
We disrupt the status quo in the lending industry by providing lenders with customer decisioning, credit risk consulting and advanced analytics to level the playing field, promote financial inclusion and support a new generation of financial products.
