By Steve Albert
Imagine it’s 1955, and you’re a loan officer at a local bank. When someone comes in for a new loan or line of credit, you sit down and get to know them. The applicant has a checking and savings account with your bank, so you know their assets. You know where they work, and you can verify their income with a paystub (or the deposits you see coming into the checking account every two weeks). They provide mortgage information and any other debts. Moreover, you may even know their spouse, where their kids go to school, and maybe the street they grew up on.
Now, imagine a company walks in and tells you they have a new, better way for you to make a loan using a model—it’s called a “credit score.” You won’t need to check all that information about the applicant. You won’t need to meet them necessarily. The applicant just provides personal information, and you receive a score that, often, immediately qualifies (or disqualifies) them for that loan or line of credit. And get this—the score does NOT take into account whether they have a job, income, savings, or lifestyle. Instead, it’s based solely on the applicant's loans and historical payments.
This sounds too good to be true. Of course, you can’t make a lending decision without those critical pieces of information that you’ve used for years, if not decades, to make credit decisions.
Or can you?
Then you think about it. If someone has a good record of paying all their loans on time, then they must have income and savings. They are very likely employed. The fact that they have a spotless record in making payments says a lot about their character—do you really need to know the applicant personally to evaluate whether they're a good bet? After all, behavioral economics tells us that we’re not that good at judging people anyway.
The commercial real estate industry is at the beginning of a similar journey.
The example is certainly oversimplified, but also instructive.
Historically, investors and lenders in commercial real estate have gathered and evaluated a tremendous amount of varied and detailed information to assess the value and concurrent “health” of a property. Even for a small building, it takes dozens of hours from investors, lenders, and appraisers to come up with an estimate of fair market value. This manual approach—of painstakingly acquiring the information to assess relative property value and health—is not dissimilar to what a 1950’s loan officer did in the days before credit scores.
While there are cases where manual efforts are warranted, the industry is realizing that in other cases, this effort is not worth the time. For a large proportion of properties, using advanced analytic approaches, valuation models provide very accurate assessments of value in a few seconds. Line items on the income statements can be analyzed and benchmarked instantly. Lenders thus can rely on automation much more than is currently the case. This is similar to credit scores, which, since their introduction, provided instant and accurate assessments of a buyer’s likelihood to repay a loan, which created a much more frictionless mortgage market.
Just as credit scores reduced friction in the personal loan market, automated valuations reduce friction in the commercial loan market.
With accurate and instant assessments of value, commercial lenders can streamline the CRE underwriting process in new ways. For instance, imagine a prospective buyer receiving a pre-approval on a loan for a property based solely on the prospective buyer’s commercial credit score combined with the property’s automated valuation. Automated values could also create new conditional approvals and thresholds. Imagine a pre-approved buyer securing a lower rate on a loan by conforming to a lower LTV (Loan to Value) ratio based on the automated valuation model.
Does a model include every piece of information typically gathered today? Certainly not. Valuation models typically do not take into account whether a building has a pool or the type of countertops. But the model does account for income, rents in the local area, proximity to schools, restaurants, transits, employment rates, and more. Similar to how payment history indicates the applicant has a job or savings, the income generated by a property really captures the presence of a pool, the fancy countertops, and the bedroom mix. The local area characteristics are indicative of rents and demand. Valuation models are based on a nationwide database of comps, which takes into account cap rate trends, and the broader cost of capital for real estate investments.
Of course, there are big differences between making a loan to a consumer and making a loan backed by a commercial building. Valuation models today do not take into account every nuance regarding the subject property itself—it’s not a replacement for an appraisal today. But it could replace a Broker Opinion of Value (BOV). And, it’s a useful tool for screening and quoting when you don’t have an appraisal, yet. It’s useful for portfolio and risk management because it’s easier to conduct more frequent monitoring of the financial health of a property.
Over time, like the evolution of credit scores, models for commercial real estate will get better. They’ll take into account more and more data about the property and the local area. As they improve, the industry will make better decisions with less wasted time and effort. Loans will close faster, and risk levels should be reduced. The impact on the health of our financial system could be immense. We’re excited to work with the industry on this journey towards improved efficiency and increased risk-adjusted returns!