Article originally published in Mortgage Introducer November 2022 – page 18
According to Douglas Adams’s 1979 science fiction novel The Hitchhiker’s Guide to the Galaxy, the answer to life, the universe and everything is 42. How did they work it out? The supercomputer told them.
Adams wrote his book more than 40 years ago, long before personal computers were mainstream – but not before algorithms.
But back to the housing market in 2022. My point is that Adams was no stranger to the idea that what computers spit out of their algorithms depends entirely on what is put in. Putting in data describing life, the universe and everything is by virtue of the volume of data, meaningless.
Considering the concepts of big data and analytics have been around since the mid 2000s, we’ve still not mastered how to use it effectively all the time. Data isn’t helpful unless you know what to do with it. Key to getting it right is not to pile in more data sets, it is to learn how to interpret multiple sources to support human judgement.
Ultimately, lenders are running a business. Their business is to lend money secured on homes that their customers want to buy. Data can be used to inform the lender how much is safe to lend against that home or to that person, but it should never be relied upon without scrutiny.
When it comes to assessing risk, which ultimately is what algorithms aim to aid, judgement must be employed because almost all data is imperfect. Factors that affect the value of a property are numerous.
Is the property leasehold? What are the ground rent conditions? What is its proximity to HS2? Is the imminent construction of a four-lane motorway planned in view of the property? Is it located on a flood plain? Is it located on what might become a flood plain given climate change? Is it too close to a coast at risk of erosion?
There are scores of data on this type of thing, and it is useful up to a point. Of course, no lender wants to approve a mortgage on a property built on top of a nuclear waste disposal site.
But it can also swing risk assessment far too far away from what is sensible.
Take London and the south east of England. The soil on which all homes and properties are built have their foundations in London clay. Clay is notoriously vulnerable to shrinkage, caused by variation in moisture content. Cutting down a tree whose roots retract can tip an entire terrace in to a twist. Climate change creating hotter, dryer spells followed by torrential rain plays havoc with London clay. Subsidence in the capital is more common than anywhere else in the UK.
So, higher risk. And as our summers get hotter and our winters wetter, that risk is getting higher still. The geological data says don’t lend. But it’s London. Jobs are there. Theatres are there. Art galleries, people, the international rich and elite are there. England’s entire transport and energy infrastructure is designed to connect London.
This is a perfect example of why it’s so important to involve judgement in decision making. Data is vital to inform good judgement but it cannot, for the moment at least, be relied upon in isolation.
Data is by its nature backwards looking as well. The past is often a very good guide to the future, but not always. I’m not suggesting that people can get out their crystal balls and accurately predict the next economic crash either, but we can recognise that change is coming and from where before it shows up in the data.
Fifteen years ago in the UK, affordability data wasn’t considered. Only income ratios and self-declared income at that. Automated valuation models used in 2006 and 2007, when house prices had been rising consistently since the mid 1990s, didn’t have the data to anticipate the crash that followed Northern Rock’s collapse and the effect that uncurbed sub-prime lending would have on property values around the world.
Anyone could see the writing on the wall by the autumn of 2007. Persistent insistence that the US housing market sneezing really didn’t mean the UK would catch a cold was a marketing exercise born of desperation as credit markets closed.
There were people who foresaw the economic catastrophe that no job, no income, no assets loans would precipitate far, far earlier.
This is why it is so important that markets remember to evolve their thinking. When it comes down to it, markets consist of people making decisions. Algorithms account for an increasingly large proportion of those decisions, particularly in stock markets. But when it comes to risk decisions in niche areas, when the law of averages does not apply, data can take you only so far.
Its value is not in its presence. It is in the prescience it can afford if given to the right person to interpret.