What are the key components of a Fintech lending toolkit?

In last month’s blog, I discussed the importance of using a dedicated full-featured decision engine (DE) for all of your credit assessment requirements. I also described why a decision engine adds significant value to a lender, beyond just using the tools that are built into a typical loan origination system (LOS).

This month, I want to expand this further and provide a high-level overview of other components that should be available within the application processing and assessment realm of credit providers. I have covered two of these components in last month’s article (the DE and the LOS), now let’s have a look at all of the others.

The components required in a best practice credit decisioning environment can be broken down into the following broad categories:

• Data Sources – these are sources of information that will be used in the application decision, both internal to the organisation, and external providers. The broader the sources of data, the broader the application pool can be. Diverse data sources also lead to stronger models and potentially greater model stability.
• Analysis Tools – these tools are used to store credit application data and analyse it for the tracking of performance and development of improved strategies and scorecards.
• Execution Tools – these components control the runtime assessment of applications. They determine the data requirements for assessing an application, retrieve and assemble the data from diverse sources, and process it through the required decisioning constructs.

As every credit granting organisation is different, and the market they are lending to varies, they may use different tools within each category, according to their requirements. A full view containing all the tools that I am going to describe can be seen in the interaction diagram below.

Let’s have a look at what the different tools are, and what value they will bring to the organisation.

Data Sources

• Application Channel – As the entry point of the application into the origination process, the application channel has the responsibility of assembling all the data that is required for decisioning. Application channels can exist in a variety of forms, from mobile applications, and self-service websites, to capture screens that are completed by telesales operators.

A good application channel balances the need to gather as much data about the customer as possible, with ease of approach that will not discourage the applicant from completing the application. With the upsurge of third-party data sources that are available now, a lot of the information that was traditionally gathered from the applicant can now be retrieved from third-party providers, some of which are covered below.

• KYC Biometrics Validation – Typically this is offered as a service by credit bureaux or government identity registers. They enable the subscriber to validate captured photographs or fingerprints against those stored by the government for official purposes.

This is a very strong KYC tool that can be used to combat third-party fraud. KYC providers can often provide enrichment data as part of their service, including contactability information for the applicant. This data can be used to either populate the application or validate information already captured. The value of this data will vary by market and provider, depending on how often and through what sources the data is refreshed.

• Income Verification Service – One of the most difficult tasks for a credit provider to perform is the verification of an applicant’s income. This can be done using submitted bank statements and/or employment contracts, but these are onerous on the applicant, open to fraud, and difficult to digitally process.

Income verification services are offered through open banking applications that enable subscribers to access information on current accounts, where salaries are typically paid. These services are incredibly useful for credit providers, but they are currently only available in certain markets and are reliant on the support of banks within the country. As a result, open banking may not be available in the market in which you operate, or there may only be coverage for some of the banking institutions.

• Credit Data Provider – Credit bureaux are the hosts of a broad range of credit data, across all providers operating in the country. They are crucial to credit risk and affordability assessment as they provide credit grantors with a complete view of the recent credit history of a customer.

Payment performance on existing accounts will inform how the customer will perform on any new credit that they are granted, and by summing up the monthly repayments of all other credit products, credit grantors can more accurately assess the disposable income that a customer will have available to service a new credit product.

• Non-Traditional Data Provider – While credit bureau data is very powerful for the assessment of risk, it is not helpful at all for applicants that are credit unaware i.e. do not currently hold any credit products. Historically, the only way to assess these customers for risk has been to use geodemographic data and application scorecards.

However, with the recent transformation of human behaviour through digital access and activity (mobile phones, social media etc.), many non-traditional data points can now be used to quantify risk. Non-traditional data providers collate this data and use machine learning algorithms to predict risk behaviour from it.

These models (and sometimes the raw data) can be incorporated into credit granting strategies hosted within the decision engine. For organisations that are operating in a market that has a large unbanked or underbanked population, non-traditional data providers are often the best source of data for quantifying risk.

Analysis Tools

• Reporting and Analytics Datamart – A toolset that stores all the information regarding applications and subsequent account performance (for approved and taken up applications). A well-structured datamart will allow for the overlay of reporting and analysis tools, as well as extracting data in structured layouts for feeding into the model development software.

A good datamart is the backbone of the credit risk department, as it enables analysts to access and extract point-in-time data along with ongoing performance for the development of reporting and credit models. A good datamart is comprehensive, well structured, and can expand as the organisation’s needs change over time.

• Reporting Interrogation Tool – Reporting tools are overlaid on top of the Reporting and Analytics Datamart for the creation of application dashboards, standard reports, and ad-hoc analysis for measuring application scorecard and strategy performance.

There are a great many reporting tools available in the market, many of which will provide easy visualisation of data by business users. Good reporting tools are flexible, can incorporate a wide variety of data, and enable users to template, refresh and share common reports through the business.

• Model Development Software – Automating credit granting decisions requires robust risk assessment models against which all applicants are processed. Traditionally, these models or scorecards have been developed by third-party vendors on behalf of the credit grantor.

Modern credit assessment is continually evolving, as are model development techniques and tools. As organisations gain access to more and richer data sources, tools have evolved to develop models quicker, using ever-expanding datasets. These models are often built for peak predictive capability, sacrificing long-term stability. Using this approach, models are continually redeveloped, using machine learning algorithms.

The data scientist will use the model development software to periodically release updated models for risk assessment. These are then deployed into the application environment via the decision engine.

Execution Tools

• Loan Origination System (LOS) – This is the primary interface between the credit provider and potential customers. The LOS is the system that allows for the generation and tracking of applications, starting with the capturing of application data, issuance of application numbers, and compilation of data.

A LOS often includes the full workflow (although workflow systems can be acquired independently). Loan origination systems should enable queuing of applications for underwriting and interface with the decision engine (either natively or through an orchestration layer).

For more information on the interaction between the LOS and the decision engine, please refer to my blog from July 2022.

• Data Orchestration System – In some organisations, data orchestration is handled independently from the LOS. A data orchestration system is particularly useful when there are a large variety of data sources required for application processing.

The orchestration system can use APIs to retrieve data from a variety of internal or external services and format this data before submitting it to the decision engine. The orchestration system should also feed all retrieved data into the reporting and analytics datamart for future use in reporting and model development.

• Decision Engine – In an earlier blog, I wrote at length about the roles of a dedicated decision engine in credit application assessment. I recommend referring to this blog for the details.

To summarise though, a decision engine provides the business with the ability to configure all elements of decisioning. Business users should be able to configure policy rules, scorecards, and decisioning strategies that can determine, amongst other things, the following about the applicant:

• Disposable income
• Risk profile
• Appropriate credit facility
• Applicable payment terms and pricing

The decision engine should support randomly assigned A/B Testing, to ensure that credit granting strategies can be continually improved.

• Document Upload and Storage – In most countries, there is a requirement on the credit lender to store documents received from the customer that can confirm income and identity. A solution is required to store these images and link them to each application.

Document management systems allow for the uploading of bank statements, employment contracts, identity documents and other supporting documentation, and links the appended documentation so that the credit team can review it as part of a manual assessment process.

The document management system should archive the uploaded files, linking them to the correct application, so that they can be retrieved in the future should an audit be required on the application.

Summary

The list of tools that I have covered in this blog is quite extensive but is not intended to be comprehensive. As the credit environment changes, new tools and services continually become available to organisations, providing them with the means to expand their credit offerings into a broader base safely, with quantified risk.

Credit application assessment is a continually evolving science, and a credit risk team should always be on the lookout for cost effective tools that can improve the credit assessment process.

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.