Prior to 2008, mortgage lending relied very heavily on algorithms – which would have worked beautifully if ONLY home values had gone straight up forever and if interest rates fell forever.
But, oddly enough, that did not happen, and the entire financial system collapsed, resulting in the worst worldwide recession since the 1930s (I hate it when that happens).
Why Did Algorithm Lending Fail So Badly Prior to 2008?
Well, besides the fact that the government created an artificial market for ridiculously risky loans (allowing rating agencies to give them AAA ratings, and encouraging Fannie Mae to buy them), the algorithms were woefully inadequate.
They had far too few data points, and the data they relied on did not go back far enough in time to properly account for the very boom and bust cycles the algorithm lending helped foster (and also home values didn’t shoot straight up forever…).
Enter my neighbor’s consumer lending firm, which relies 100% on algorithms to make loan decisions.
Sidebar: Algorithm lending relies entirely on automated systems to make credit decisions based on multiple data points in lieu of traditional underwriting that relies on humans to review income, credit, and asset documents.
My neighbor’s firm relies on OVER 2,500 data points to make consumer loans. And despite the clear economic downturn we saw in middle America over the last few years and are seeing now, they have NOT seen a significant increase in defaults.
Their algorithm is so powerful, in fact, that they saw fewer defaults than expected.
The data points are accessed automatically and include non-traditional info such as college majors, GPAs, SAT scores, social media engagement, professional certifications, bank account transaction patterns, and even device usage patterns.
Algorithm Lending’s Returning to the Mortgage World
All this is to say that algorithm lending is now far more sophisticated and accurate – and it is returning to the mortgage world, thankfully.
And yes, the mortgage world continued to use algorithms in conjunction with old-school underwriting after the 2008 crisis (e.g. Fannie Mae’s Desktop Underwriter), but it is not nearly as sophisticated as my neighbor’s firm’s model and lenders still force humans to review files so they can put all too many borrowers through documentation-hell.
Politicians and mortgage regulators frown on algorithm lending too, because it can seem to favor the wrong demographics and/or remove regulatory control from politicians.
5-Minute HELOCs Are Here
Despite that, we now have two Home Equity Line Of Credit (HELOC) lenders who rely entirely on algorithms – much like my neighbor’s firm – and it is glorious!
It is glorious because it takes about 5 minutes for us to get an approval, and borrowers can close in as quickly as five days! And we don’t have to run borrowers through the old-fashioned underwriting gauntlet (no documents needed).
And given that we are seeing regulations loosen up in the mortgage industry finally, we will hopefully see full algorithm lending return to the realm of first mortgages.
When that happens, borrowers really will be able to finally “push button, get mortgage…”
