Business is booming in niche and non-conforming mortgage markets, and lenders in this sector have seen their business increase significantly in the last few years. At the same time a number of lenders have increased their product ranges to include sub-prime and flexible products, second charge loans and bridging loans, mirroring the actions of intermediaries who are experiencing similar growth and diversification.
Throughout this period of rapid growth and change a core of non-conforming lenders have steadfastly held to the view that the individual underwriting of cases is the only decision-making process that can work for the non-conforming market, and the vast majority of intermediaries operating in this market sector appear to share this view.
One powerful reason supporting the underwriting route for decision-making is the ‘service factor.’ Nevertheless, lenders can no longer afford to totally reject the idea of using automated selection processes, because judicious use of at least some level of automation could help lenders ‘ and by definition brokers ‘ to continue to provide and enjoy top quality service, while still giving cases the personal attention they need.
To understand the automation choices now facing the non-conforming mortgage sector, including credit scoring, it is important to look at the historical reasons for favouring individual underwriting. First, a high proportion of non-conforming applicants have already been declined by credit scoring systems, and have circumstances that need individual consideration.
Second, the decision to lend must be a holistic one, taking into account the underlying causes of the applicant’s financial circumstances. The applicant must be assessed on their ability to make the required payments regularly and on time ‘ and this can only be done by building up an overall picture. Niche underwriters are very experienced in doing this.
Third, advisers appear to prefer underwriting on niche business as it enables them to explain why an application has been declined ‘ something that credit scoring cannot deliver.
Finally, non-conforming lending is relatively new in the UK, and volumes are relatively small compared with the mass market. So far, lenders have not amassed the huge quantities of borrower information needed to construct effective credit scoring.
Clearly non-conforming applicants are the sort of people who are very hard to fit into pigeon-holes. However, with enough data available, a credit scoring system could be devised for non-conforming lending. And although it probably could not account for 100% of the decision-making, it could achieve considerable efficiencies. These could be achieved principally by a credit scoring system that deals with the best applications (those that are well within criteria and get immediately accepted by the system) and the worst ones (those that are declined straight away, enabling the broker to quickly place the case somewhere else). This leaves the mid-range applications to be underwritten.
A number of lenders have made a virtue of the fact that they do not credit score ‘ and this gives a great deal of comfort to advisers operating in the non-conforming sector. However, many lenders are now standing at the crossroads. At a time when former non-conforming specialists are looking to expand quickly into the mass market, how long can they keep on making lending decisions in the same way across a diverse product range? And perhaps more importantly how can they achieve a greater degree of automated application screening and avoid their advisers thinking that they have ‘sold out’?
Most lenders remain rooted in the idea that using full credit scoring to make lending decisions on non-conforming mortgages is not the way forward because there are always individual factors that credit scoring cannot cater for, but they must be alive to the fact that even partial scoring might produce business efficiencies for them and their advisers ‘ especially when they consider the wider market they now operate in: prime, flexible, second charges, and bridging loans.
At the moment, the potential options fall into three categories: the use of scoring based on existing credit reference data; generic credit scorecards; and the creation of a made to measure scorecard which is specific to a lender’s own business.
Credit reference data scoring is a limited form of credit scoring that has been available for many years. It is based solely on information in the applicant’s credit reference file that is known to have a bearing on credit performance. For example:
• The number of addresses is indicative of financial stability
• The post code can indicate the likelihood of default
• The performance on other financial transactions is a pointer to how the new mortgage loan will perform
• The score will factor in any adverse data such as CCJs and/or bankruptcy
Using these indicators, the credit reference data score will be able to give an estimate of how good a credit risk the individual is. However, it does not include certain key factors that the lender will need to be aware of in making the lending decision. For example:
• The mortgage payment record
• The applicant’s job stability
• The quality of the property
In this instance, all of these are still the domain of the underwriter.
The second option is to use a generic scorecard. Credit reference agencies have pooled their experience of individual lenders to produce generic scorecards, but how far are these useful to niche lenders? Although they are based on very large quantities of data, they are still pooled experience, and not based on lenders’ own lending experience. And as with the first type, these scorecards cannot do what the underwriter does ‘ take into account:
• Employment types
• Income stability
• Whether the loan is for purchase or remortgage
• The property itself
From the point of view of traditional non-conforming lending these scorecards cannot supply generic information on the non-conforming sector. However, for prime business, one way forward would be to start with a generic scorecard and build on that to create a bespoke system.
The third option is to create a scorecard from scratch. This is the sort of credit scoring that most people recognise, and which has come to have an aura of ‘mystery’ and ‘inflexibility’ around it ‘ especially if the application has been declined. It is based on the lender’s own data on the sort of applicant that results in a well performing loan. Larger lenders can take their own bulk data to create their credit scorecards. These will be specific to the style and experience of the particular lender, and can be easily adapted in the light of more current data as it becomes available.
However, not to be daunted by these apparent difficulties, some non-conforming lenders have already started to develop ways to make a more automated approach work for their own business. Second charge loans demand a very quick decision ‘ so they are the obvious choice to start looking at a semi-automated route into decision-making, such as a decision tree on a website. This is a good methodology for low value/high volume business such as second charges, where there is less data required from the borrower than in first charge business, and consequently not so much documentation on which the case could fail.
So can this decision-tree process be refined and backed into lending on first charges and on prime business? One thing is for sure: lenders cannot afford to stand still and they constantly need to keep reviewing all the options. For example, some are currently running tests to see if credit reference scoring and indebtedness scoring systems offered by Experian would have been predictive of the performance of their own individual loans. In the USA, credit scoring is already widely used by sub-prime lenders to make effective lending decisions, and many speculate on whether it might be brought across the Atlantic. The problem in the UK is for it to get started. No one wants to invest in new systems until there is a track record of success. Lenders currently investigating automation must be convinced that it will be an improvement and will be welcomed by mortgage advisers.
From a business point of view, if lenders and brokers end up rejecting any sort of automation, then they must have some very clear reasons why they chose that path. As a lender, some level of automation might well make our decision-making processes more efficient ‘ but we can never afford to forget what gives the intermediary comfort, which is a personal, fast service delivered by reasonable and accessible human beings.
A degree of automation can speed up the process while retaining the service that cases need.
Non- conforming lenders have three options: existing credit reference data, generic scorecards or create their own.
Non-conforming lenders are waiting for someone to make the first move before introducing credit scoring in the UK.