The Bank of Canada recently rained on hopes for a quick recovery from the economic impacts of the country’s coronavirus slowdown — a hope nurtured, many thought, by signs of economic resilience over the summer.
Alas, late in October, the Canadian central bank issued a Monetary Policy Report suggesting that hard times will continue, perhaps all the way through 2022.
“The Bank estimates that over 2020–23, persistent scarring effects of the pandemic on the labour force,” the Bank of Canada writes. The word “scarring” is a favorite of chief central banker Tiff Macklem, who has been issuing periodic warnings of tissue damage to the Canadian economy since the pandemic was declared in mid-March 2020.
Assessments like this should prompt non-traditional lenders — conceivably any business that might want to extend credit, from retailers to car dealers — to rethink how they gauge loan applicants’ creditworthiness. Why? Because businesses of all sizes in every province and territory will be looking for capital to help them through a recession made worse by:
- A pandemic that isn’t over, resulting in
- The need for ongoing social-distancing measures that can snag business recovery, such as expensive new workplace configurations and equipment
- Pinched household budgets and other recession-related woes, leading to
- Subdued consumer spending
Meanwhile, many lenders will be making credit decisions based on inputs that aren’t adequate to the times.
CREDIT SCORING IN CANADA IN THE FACE OF UNCERTAINTY
Applying old-school analysis where new market conditions prevail could curtail lending and stall economic activity. For lenders of this ilk, being behind the times could jeopardize opportunities to make sound and profitable loans.
Lenders traditionally rely on credit scoring based on objective financial data and subjective views on some of the would-be borrower’s non-financial traits. In this approach, the financial data includes the prospective borrower’s credit history, and line-item comparisons of historical financial statements submitted by the applicant. Among traditional non-financial inputs are the would-be borrower’s profile (employment, status, degrees, home and car ownership), a qualitative assessment of the borrower’s previous dealings with the lender, and, for business loans, relevant business plans.
In this approach, financial data has more weight in determining how stable and efficient the applicant is when it comes to their finances.
Although recent word from drug maker Pfizer seems to bode well for a Covid-19 vaccine, the company’s claims have not been verified, and, given the logistic challenges, widespread distribution of a vaccine is unlikely before, at the earliest, mid 2021. For now, it’s prudent to remember we don’t actually know how long either the pandemic or its economic aftershocks will last.
“These uncertainties erode the rationale behind applying only traditional credit analysis,” says Elena Ionenko, co-founder and business-development head of lending-technology provider TurnKey Lender. “After all, real-time financial data can be as indicative of repayment as historical information.”
Adds Ionenko: “This analysis can be performed on a continuous basis — triggering monthly or quarterly reports — that provide dynamic updates on the loan, which helps lenders see how the borrower is coping in real time with the challenges of a recession, while comparing these results to pre-crisis data.”
GUIDED DISCRETION
In a typical lending scenario, lenders start off by “scoring” loan applicants to determine the likelihood of their returning an amount owed with interest in a given period. Most use third parties such as Fair Isaac, whose Canadian FICO scores assign numerical values between 300 and 900, with 900 indicating maximum creditworthiness.
For the most part, this traditional scoring relies on factors such as:
- How long the applicant has been using credit
- The amount and type of debt an applicant already has
- Current interest rates on outstanding accounts
Lenders use these reports to generate a risk profile of the applicant, which helps lenders determine whether to make a loan in the first place, and the terms of any loan that’s approved. Obviously, applicants with low credit scores tend to be assigned higher rates of interest than those with higher scores, though the ultimate decision is made by the lender, with FICO inputs used as guardrails.
Of course, the pandemic has eroded the credit standing of many who have lost jobs or seen wages cut, necessitating new ways to evaluate consumer creditworthiness.
LOW FICOS AND THE UNBANKED
For example, a FICO score won’t tell if an applicant has lost her job or seen her income dip in the public-health crisis. One solution to this increasingly widespread problem is working with alternative data sources for determining creditworthiness.
One of the most reliable sources of information? An applicant’s bank accounts.
Some lending-technology providers empower lenders to examine applicants’ bank accounts and track transactions to take note of spending habits and monitor employment and non-employment income including such responses to the pandemic as stimulus payments, forgivable loans, and unemployment-insurance proceeds.
Some advanced lending-tech firms equip lenders to see these data points, and more. For example, alternative scoring can uncover normally hidden risks such as an applicant’s gambling expenditures and overdraft durations and apply them to credit decision making.
And for consumers who are unbanked or underbanked — 18% of Canadians, according to ACORN Canada — alternative scoring is a must. More so when you take account of LexisNexis research indicating that 51% of traditionally unscorable applicants in the US are as creditworthy as consumers with high traditional credit scores.
AUGMENTATION, NOT REPLACEMENT
“This doesn’t devalue traditional credit scoring,” says TurnKey Lender’s Ionenko. “For predictive power, no one alternative approach is as formidable as credit-bureau input.” Alternative data points are more numerous, more scattered, and less organized than the data that contributes to a traditional credit score, she explains. “This means neural networks and other AI-based tools are required, which is an approach we pioneered.”
Fortunately, these resources are now available to lenders, and the additional intelligence this normalized alternative data provides helps lenders understand their customers better, make better loans, and build better-performing loan portfolios.
Years using alternative data sources to assess consumer creditworthiness in the developing world has paid off for some lending-tech makers, giving them a solid sense of a loan applicant’s relative riskiness. As a consequence, lenders plugged into such technologies in developed economies have new ways to gauge the “lendability” of consumers — and new ways to contribute to the economic recovery of markets in Canada and around the world.
In particular, fintechs know that bank-statement data, enhanced by deep machine-learning and time-tested artificial intelligence, means better outcomes for borrowers and lenders alike.