We have purchased mortgage assets from a number of specialist lenders in the past, some of which use automated decision tools and some of which underwrite manually. In my experience, automated underwriting produces demonstrably superior results compared with the manual variety.
Stephen Knight, executive chairman of Checkmate Mortgages, wrote a letter to this magazine a couple of weeks ago explaining why this is the case and I agree wholeheartedly with him (Mortgage Strategy March 24).
Of course, it can be argued that automation causes as many problems as it solves. The US market has the highest penetration of automated underwriting technology in the world and unfortunately for the rest of us, it was the US that processed sufficiently crass credit decisions to bring the global economy to a halt.
How did this happen? Put simply, the main benefit of automating underwriting decisions is that the process is not under pressure from sales teams, brokers or the need to hit targets. Neither is it subjective, prone to errors or able to be co-opted by fraudsters.
But machines operate within parameters set by humans who, of course, are prone to various pressures. If you programme rules into an underwriting engine that allow 150% LTV loans up to £500,000 at base rate minus 1% to borrowers under the age of seven who are self-certifying their pocket money of £120 per year, your machine will grant those loans. Even the most relaxed securitisation investors might realise that such loans are not a good idea.
US sub-prime lenders were guilty of setting their decision engines to accept loans that should have failed. The efficiency of their front-end processes meant they were able to get things more wrong more quickly than a manual underwriting approach ever could. Meanwhile, securitisation investors were too greedy to notice, allowing lenders to perpetuate mistakes.
If and when the sub-prime market returns in the UK there will be an intense focus on the consistency of underwriting and credit controls. These factors will be measured against the performance of assets. Only automated systems can generate the required level of control and all surviving lenders will need underwriting technology.
Some lending practices that have prevailed until recently may not survive. For example, self-cert deals will be hard to fund as risk will have to be measured accurately on each loan. This is impossible without a full assessment of income.
So automated underwriting is the future but the human rules guiding it are going to get much tougher.