The next step for automation in banking
As the demand for SMB lending grows, banks can use automation to capitalize on an underserved market.
By
As online banking has grown, banks have met the expectations of digitally savvy customers who prefer to open their accounts online by automating much of the account-opening and customer-onboarding process.
While banks have been rushing to keep up with the digital revolution, another chasm has been building in the background for the past 30 years.
In 1994, there were 10,329 community banks. By 2014, that number shrunk to 6,094. From 1994 to 2018, community banks’ share of banking assets and lending markets fell by 40%. In 2020, community banks accounted for just 15% of the banking industry’s total loans.
Causes for the decline include the savings and loan crisis of the 1980s and the 2008 financial crisis, as well as increasing regulatory and compliance costs and the removal of barriers to bank consolidation.
Why does this matter? Community banks are typically the largest source of funding for small rural communities, and they hold 36% of small business loans. Since community banks are able to form personal relationships with business owners, they’re able to make risk assessments based on “soft factors,” rather than just the hard metrics that larger institutions rely on. This makes them better able to finance SMBs.
As the footprint of community banks shrinks, SMBs have had to turn to larger banks, only to find them less willing to provide loans to them as the banks have become more cost-conscious. The cost of underwriting a loan is the same, regardless of loan size, making it uneconomical to fund smaller loan amounts.
At the same time, the recent COVID-19 pandemic accelerated digital adoption, making business owners more prone to apply for a loan via phone, rather than walk into a physical branch.
These three factors have combined to create a huge unmet need for small business lending that is waiting to be filled. The key for banks to capitalize on this opportunity is automation.
Automation use cases in lending
Current processes for pre-qualifying and underwriting loans are time-consuming and manual, leaving banks with limited staffing resources to expend precious attention on small business loans, instead of focusing on larger loans that would yield higher individual returns.
Additionally, credit scores are less effective in predicting borrower performance for small businesses because small businesses often have a limited credit history, requiring decisioning on the owner. However, this tells a lender little about the business itself.
Transaction analysis is a much better leading indicator of business health. Using data science, banks can utilize transaction data and cash flows to measure the health of the business and prequalify borrowers before a human even gets involved.
Transaction analysis
In the case of assessing a bank’s existing depositors, data is typically sourced directly from the bank’s core system. However, the same approach can be used for customers of third-party banks. In such cases, the borrower may submit bank statements or log in through an aggregator like Finicity, Plaid, or Yodlee. Optical character recognition (ORC) technology is used to ingest statements and turn them into machine-readable text that can be assessed similarly to aggregated or bank-sourced data.
Once the data is ingested, the system will analyze the data through the following steps:
- Classify transactions (revenue-based deposit, non-revenue deposit, NSF, third-party debt service, or suspicious transactions).
- Tabulate results to generate a monthly, quarterly, or annualized profile.
- Run that profile against a transaction policy (maximum NSFs, minimum number of deposits).
- Establish the pricing/offer amount based on annual gross revenue, credit score, and other factors.
- Generate prequalification offers.
Prequalification
Regional and community lenders are often challenged with what, how, and when to market lending products to their customers and members. Many report an email or direct mail response rate of roughly 1%. Even once the potential borrower takes action, the approval rates are meager—typically below 40% but often substantially lower, based on Lendio’s experience with our broad base of lenders.
Transaction data can be used to generate a pre-qualified offer that the lender can stand behind. Imagine the level of engagement for a business owner as they receive a message that says, “Come apply for a loan,” compared to, “John Doe, LLC, you’ve been pre-qualified for a loan of up to $75k.” The use of transaction data, accompanied by personalization of the marketing message, significantly improves the effectiveness of marketing investment and lender teams.
Underwriting
Once a customer has accepted the pre-qualified offer, he/she is prompted to verify some demographic and firmographic information. In parallel, to collect that information, the system will verify the information and pull additional third-party data including:
- Credit bureau checks
- KYB compliance checks
- Identity verification
- Tax documents
- Account verification
Once the borrower has been evaluated, the system can finalize approval within 15 seconds of receiving the application. At the same time, lenders can choose the level of automation they would like to utilize. That may range from automation of application ingestion and data pulls with a human review to a fully-automated approval or decline.
Benefits of banking automation
The automation of banking systems has benefits across the board including:
Customer experience
Customers want a streamlined, online experience that can get them an answer fast. With automated checks from multiple data sources, banks can provide instant pre-qualification based on robust criteria.
Regulatory compliance
By building and using automation systems built to match the bank’s risk and compliance policies, banks can have more confidence in their decision, instead of resorting solely to shoring up loan offers with higher rates.
Cost savings
The lending process immediately becomes less time-consuming and expensive when all of the data that a human used to sort through is automatically tabulated by a computer. This frees up staff time to focus on areas that require human judgment.
Predict better outcomes
The Bank for International Settlements found that alternative data could predict future loan performance better than traditional methods, especially in areas with high unemployment.
Automation challenges
Of course, automation comes with its own set of challenges.
Outdated core banking systems
Many banks use outdated core systems that are notoriously difficult to work with. Any automation technology will need to be built with the flexibility to work with a less friendly system.
Managing the technology
Banks don’t necessarily have the resources to manage and apply the technology independently. Look for a SaaS platform maintained by a third party that’s API-ready and can be configured to varying product parameters and pricing models.
Fraud concerns
Combating fraud is a crucial component of adopting automation. Putting safeguards in place through KYB and IDV verification is critical to balance risk with the benefits of automation. Additionally, by targeting existing depositors for prequalification, banks can work with a pool of already-vetted accounts.
Customized loan products
Certain loan products, such as commercial real estate, require much more customization to the borrower, making them harder to automate. Typically, the same thought processes have constrained automation in SMB lending. However, with new technologies, borrower risk can be more easily assessed thus enabling far greater levels of automation.
An untapped opportunity and improved accessibility
Banks that effectively adopt automation to overcome small business lending challenges will be able to access an untapped market. But the impact goes far beyond a substantial revenue and cost savings opportunity.
Underserved communities, such as those with higher immigration and crime rates, are frequently overlooked because of the inherent biases in credit scores. By relying on alternative data sources such as transaction data, banks can get a better understanding of the business’ health and create fairer outcomes for those communities.
Interested in lending automation? Learn more in this ebook, Automated Lending: A Mandatory Upgrade.
