In my previous post, I wrote to you about Zopa’s performance during the 2007/8 recession. This time, I wanted to go into more detail about how we use our decade’s worth of data in risk analytics.
11 years of data
In risk analytics, the length and relevance of the data that you have to analyse is extremely important. Our 11-year history puts us in a unique position regarding this: for example, we are the only peer-to-peer lender that operated through the 2007/8 financial crisis, and so we know how our loans perform under a stress environment. We have used what we learnt from this period to inform our current underwriting strategies and set loan pricing: with the aim of protecting lenders from losing capital should a similar event occur.
Our focus on prime borrowers and a consistent product means that the data collected and the associated learnings during our 11+ years are still relevant today. It’s what allows our seasoned risk analysts to continuously hone our underwriting strategies with more than a decade’s worth of incremental improvements.
Depth of data
Since the company was founded, we have lent over £1.5 billion across 225,000+ loans. Our database contains information on every loan applied for from Zopa (including declined applications, which we anonymise): so we now have a dataset of over 2 million loans. Our expert analysts use industry-leading techniques to mine these data when building our credit models and underwriting criteria.
At the application stage, we collect thousands of pieces of information on each applicant from two credit bureaus and our application form. These data are used to evaluate the applicant across multiple dimensions:
- Income stability
- Attitude towards credit
- History of managing credit
- Ability to afford the loan
All of which helps us build a detailed and clear picture of the potential borrower. Our decision to approve a loan is based upon them passing our strict underwriting criteria.
Credit bureaus in the UK allow us to track the credit performance of declined loans through anonymised data. By tracking how the loan performs over time, we can accurately assess our decline decision without conducting hard credit searches, and to stay on the right side of data protection laws.
We are always looking for more ways to improve our underwriting strategies. To enable this, we purchase additional anonymised data from the credit bureaus to track the credit performance of our existing loans and declined applications to refine our future decision-making. At Zopa, we are constantly adding to our wealth of data and are committed to making the best use of it.
In my next post, we will be taking a closer look at Safeguard, explaining its purpose and resiliency.
Sharvan Selvam is Head of Risk at Zopa.