How data modelling is opening up funding opportunities for mid-sized SMEs
Many mid-sized SMEs have found it difficult to secure lending over the past two years as lenders have struggled to assess the true impact of the Covid pandemic on their businesses. Lenders using off-the-shelf credit ratings often fail to distinguish between the potentially stronger scores of mid-sized companies versus their smaller counterparts within the SME sector, viewing them as a homogenous group. In contrast, ThinCats’ proprietary credit risk model PRISM, built specifically for mid-sized SMEs, allowed it to provide a record £318 million in funding in 2021 alone.
“Many of these firms needed loans the most during the pandemic - to supplement uncertain income, especially in periods of lockdown - but less advanced credit models did not have the depth of information or the flexibility to cater for them,” explains Gareth Rumsey, ThinCats’ Head of Data Analytics. “Improved accuracy and precision in the latest version of PRISM has meant we can continue to provide valuable funding to this underserved segment of the market with confidence and at competitive prices.”
More than three-quarters of mid-sized population now creditworthy
PRISM uses multiple sources of financial information to assess the likelihood of default on loans of between £1m and £15m. Developed last year, using sophisticated machine learning techniques, the latest version is 27% more accurate at predicting insolvencies than the previous model. This improved predictive power increases the proportion of mid-sized borrowers that ThinCats deems creditworthy from 57% to 76%, yet the observed 4-year insolvency rate within this larger creditworthy population is just 3% - even lower than before (3.6%).
Increased transparency and improved collaboration
One of our key objectives in upgrading the model was to facilitate open communication with both potential and existing clients. PRISM can now be viewed via a credit dashboard where the positive or negative contribution to the Risk Grade of metrics in seven risk variable areas - gearing, liquidity, profitability, efficiency, growth, demographics and filing behaviour - are visually presented. This dashboard facilitates rapid understanding of areas of strength and weakness within the business and direct the ThinCats underwriting team towards anything that needs exploring in more detail, creating greater opportunities for potential clients to influence the decision-making process.
For existing clients, we can recalculate the risk grade based on monthly management accounts, to identify potential changes to circumstances and initiate meaningful conversations before the risk of default arises.
Greater confidence in indicative prices
PRISM is calculated predominantly from annual financial accounts, meaning the grade rarely changes month-on-month. Clients can therefore be confident that the indicative price quoted based on the risk grade will not change by final terms.
The model has been extensively tested on 15 years of observed insolvencies among mid-sized SMEs. This has allowed us to ensure that it works consistently throughout an economic cycle as well as across different industry sectors, sizes of business and geographies. So, we can always be confident that our pricing remains fair and appropriate.
Adapting to changing conditions
On the other hand, because the risk grade is calculated from accounts that may be up to 18 months old we needed to build in some flexibility. The latest upgrade allows us to react input alternative figures to the filed accounts - such as more recent management accounts - into the model. This ensures that we can identify cases where the latest filed accounts paint an unrealistic picture of the borrower’s financial situation. We have also added a second indicator, Spectrum, which works alongside Prism. This blends credit and current account information that banks and other lenders share with credit reference agencies into a traffic light system that flags up red or amber alerts in cases where a negative event has occurred since the latest accounts from which Prism is calculated
The PRISM model has already enabled mid-sized SMEs to borrow more than £1.2bn through ThinCats. We expect data analytics to continue to underpin the growth trajectory of our lending to this critical part of the SME market.
In recognition of this progressive approach, we are proud to be among the nominees for the ‘Best data initiative of the year's at the upcoming UK FinTech Awards 2022.
