Why are we publishing this data?
We published an annual expected loss figure in the past, but removed these figures in July as the calculation had not been updated for some time and as we felt the information was not meaningful for the RateSetter model due to the protection against defaults provided by the Provision Fund.
However, we think that this data can be useful for existing and potential investors, so we have been working to improve and update our calculation methodology and to ensure that the numbers can be updated systematically and in real-time.
What does this mean for reporting on the Provision Fund?
We already publish real-time data showing how much of the contributions into the Provision Fund in any given year have been used during that same year. This is the row of data titled "Provision Fund usage". A figure below 100% indicates that based on the latest available data, more has been paid into the Provision Fund than has been taken out to cover bad debts for that year.
We will continue to publish this data, and information on the size of the Provision Fund and Coverage Ratio, which can be found here.
What do the new figures show?
"Expected lifetime bad debt rate" is a measure of the expected level of losses on loans written during a calendar year. It takes account of losses over the lifetime of those loans even if they occur after the year the loans were written. It is calculated at the point that new loans are written and updated throughout the year. At the end of that calendar year, the figure is locked down.
"Latest projected lifetime bad debt rate" is a measure of the expected losses on loans written during a calendar year, based on the latest available data. This figure is updated on an ongoing basis, taking account of losses experienced to date, plus an updated loss forecast for loans that are still active. As time passes and the loans mature, the figure becomes more accurate.
How are the new figures calculated?
The calculation of both figures is automated and updated in real time.
When each loan is written, it is put into a risk band according to how likely we believe that loan is to repay in full.
We then take 12 months’ worth of loans from the same risk band which were written at least 18 months ago, and look at actual default rates for those loans. We also factor in the amount that we expect to lose in the event of a default (this is called "loss given default"). Using this data we can calculate the expected bad debt rate for each risk band.
This is a well-established approach across the finance industry and banking. It means that our expected and projected loss figures are calculated based on actual loss rates for real loans with similar characteristics.
The projected bad debt figures take into account the most recent data on the performance of that cohort of loans.
The process for calculating the expected bad debt rate and projected bad debt rate is more qualitative for business and property development loans: these loans are younger, there are fewer of them and defaults are rarer, so our data is less extensive. As these areas of lending grow and the loans mature, the calculation will become more quantitative.