Bondora has analyzed more than 1 billion loan applications since launching its operations in 2009 and has developed significant experience in credit scoring using the data it has amassed. With a proprietary credit scoring database that provides critical insights into borrower behavior, Bondora has developed a high level of expertise in respect to consumer lending in Estonia, Finland and Spain.
Bondora employs an internally developed credit scoring model, using all the data collected in earlier steps to calculate a prospective borrower’s risk rating. The exact variables that influence credit risk ratings are determined and adapted as needed through statistical analyses. These may include income information, employment records and credit history, as well as marketing and other non-traditional data.
Bondora Rating is calculated using all the data points we have on an individual. Among other things a lot of emphasis is on externally validated data we get from credit bureaus, population registries, banks and tax authorities. The data we get from these trusted third parties is very sensitive and cannot be shared.
The self-reported data that is displayed to investors is typically not very predictive and has relatively little influence on the Bondora Rating of a particular loan.
The individual risk of giving out this particular loan is calculated using all the parameters we have collected. Quite often, but not always, traditional metrics have not been very predictive for evaluating borrower risk and instead much more emphasis is fully verified data, especially behavioural data, from trusted third parties. That is why it is impossible to derive the statistical risk level of a loan simply from the data that is visible. A borrower with high income and few other credit lines might get a high risk Bondora Rating and a borrower with low income and many other credits might get a low risk Bondora Rating. This is because data we get from third parties may identify behaviour that is respectively high risk or low risk.
Bondora Rating represents the combined expected loss (loss after recoveries), not only the default risk of a loan. This means that when calculating the Bondora Rating we also estimate the likelihood of recovery. If two loans have equal expected default, but one has a substantially better expected recovery rate, then the ratings would differ. In other words a loan with lower default but also lower recovery potential can get a riskier Bondora Rating than one with higher default but also higher recovery potential.
Additionally, loan applications of the same borrower can have a different Bondora Rating as the potential for recoveries is different for different loan amounts.
Scorecards are prepared by Bondora’s consumer credit business unit scoring and price team and signed off on by the unit’s Head of Credit Risk. Scorecards are regularly evaluated and calibrated.
Scorecards provide one-year anticipated estimates of the expected loss (i.e., the proportion of gross interest not received due to loan credit losses). Based on this data, loan applications are assigned a Bondora rating ranging from AA (the safest grade) to HR (the riskiest “investment grade” rating).
Minimum and maximum expected losses per risk rating are provided in the following table:
Risk Rating Min EL% Max EL%