Advanced AI Redefines Calculation of Credit Risk

Risk Button on Abstract Purple Data Background

The 2008 global financial crisis woke the world up to the devastating risks inherent in the credit industry. Canadian banks remained a model of fiscal prudence in one of the largest financial downturns in history, but the legacy of the crisis endures in our country, nonetheless. Millions of Canadians rely on banking credit daily – from the small transactions on a credit card, to the decades’ long borrowing schemes propping up homes and businesses. These larger money-lending instruments play a vital role in Canada’s economic wellbeing: nearly 98% of Canadian companies are small businesses. These companies rely on bank loans to set up shop and begin playing their role in the economy.

The way large banks and financial regulators have traditionally defined risk, and therefore eligibility for funding, can neglect useful information, leading to suboptimal decisions. The current paradigm of risk assessment frequently denies small business loans to applicants who are beginning a commercial enterprise for the first time unless they have already amassed a substantial amount of wealth. For individuals with lower incomes, less wealth, or little job experience, the way risk is assessed creates a high barrier of entry to loans. Those who do receive loans often must repay them at a higher interest rate. This system both favours applicants with pre-existing wealth, thereby exacerbating a wealth gap in Canada, and results in an unnecessary stifling of small business growth.

Results

Cristián Bravo Roman and his research group in the Department of Statistical and Actuarial Sciences at Western University have developed advanced artificial intelligence methods to redefine the way we calculate credit risk. “Technology has changed leaps and bounds – computational modelling allows us to do things we never dreamed possible, yet still the criteria for determining credit risk have remained relatively static,” says Bravo Roman. His group’s statistical models take into account a wide variety of applicant information, including satellite imagery of their neighbourhood, tailored evaluations by experts, and the borrower’s financial social network to build a richer picture of credit risk than ever before. In addition to opening up the playing field, Bravo Roman’s models can make these financial decisions at high speed, avoiding the long and costly evaluation systems in place currently.

Changing traditional criteria for assessing credit would increase investment in Canada by approximately $7.5 billion without affecting the lenders’ risk. As this funding flows from banks and other lending institutions to small business owners, Canadian employment will likewise benefit; on average, for every small business funded, one additional job is created. On aggregate, three quarters of all employed persons in the country work for a small business. The profound social and economic progress engrained within our system of business loans enables the success of millions of Canadians. As technology allows for improvements to this system, millions more will be able to strengthen their livelihoods, and with them, the nation’s economic resilience.