Data is full of errors: Quality is more valuable than quantity!

Last October, I wrote about "Big Data" and how it could help lenders use analytical tools to assist in predictive risk management and marketing. This week, the American Banker magazine featured an article entitled "Data Sold to Lenders is Full of Errors." I don't feel I have egg on my face. Good data analyzed by good analytical tools will produce credible results. Unfortunately if the data is not good or the analytical tool is not good, the results will be questionable - perhaps disastrous. Lenders have great data. The data they already have is probably much better than the data they purchase. It is in multiple databases derived from loan applications, appraisals, credit reports and income & expense submissions. The quality of the data requested from banking customers can be verified and cross-checked as it is collected. A good appraisal review function, underwriting function, credit function and risk management function - properly functioning within a well-managed bank - has all the data they need. And they have better data than any data that they might purchase. Those four functions can cross check information obtained from the other functions. They will end up with information far more accurate than what they purchase. Once they realize that they already have great data, lenders should find the best way to manage and mine it. Certainly, Information System professionals will need to get involved in the organization and storage. Once that is done, the exciting part begins - analytics. The University of Massachusetts at Amherst legitimately holds claim to being the #1 research university in the field of analytics. Certainly, lenders in Massachusetts can find a recent graduate or a current graduate student capable of massaging quality data to manage risk and identify market trends - often under the watchful eye of a brilliant professor. The appraisal profession has been collecting data one property at a time for years. We collect data about the subject property from the property owner or the lender. We collect data about the comparables from publicly available sources - from real estate brokers, assessors, buyers, sellers, etc. But that data is potentially suspect depending on the provider's perspective or depending on their confidentiality constraints. Every appraiser I know would love to have the quality data that banks have available to them - if not in detail, at least in the aggregate. USPAP and Graham Leach Bliley confidentiality issues prohibit appraisers from identifying the financial details of a comparable property that were obtained as a result of an assignment. So, our reports can either disguise the true identity of the property for which we have very good data, or the report can clearly identify a property for which we have less reliable data. Either way, our client - the lender - may get a slightly clouded version of the required information. Lenders, on the other hand, have significantly higher quality data available to them than they might get from an appraisal or, more particularly, for a public or subscription data source. A bank typically gets a copy of every tenants lease, every landlord's operating expenses, the actual tax bill and copies of a borrowers's tax return. Best of all, this data is freshly available at least once a year. And if the bank's staff is alert, they are getting information on every new tenant and every lease renewal. Analysis of this data provides a reliable data bank that can be analyzed. Quality is far more valuable than quantity! For some reason, too many lenders miss the point; they might believe that their data is somehow less reliable than the data they can get from "big data" vendors. In fact, it is far more reliable. And it is also confidential. So, if they are in compliance with customer confidentiality laws and regulations, they are certainly not providing the data to the brokers of big data. And if the brokers of big data can not get it from the most reliable, most confidential source, where are they getting it and how valuable can it be? Shaun Fitzgerald is the owner of Fitzgerald Appraisals, Easton, Mass.