How can a computer value your home? Buy a First Class Home

How can a computer value your home?

Posted by Carry Brooks on October 25, 2011 Leave a comment (0) Go to comments

Its been 10 year since Veros Real Estate Solutions entered the then-emerging world of automated valuation models, where computer databases estimated a propertys value instead of a human appraiser eyeballing the property and its sales history. The industry has helped check the cost of appraisals as well as bring real-time valuations to institutions and consumers alike. AVMs have been controversial, too, with questions raised about the accuracy and fairness of the work. In light of Veros anniversay, we checked in with William King, Director of Valuation Services at the Santa Ana-based company to learn whats up in this trade.

Us: How has the automated valuation business evolved in the past decade?

William: The first generation of AVMs was introduced in the late 1990s, establishing the product as a valuation tool. The second generation of AVMs, introduced in 2002, provided updated models and additional ways to provide valuations. As the industry moves into the third generation of AVMs, new data, data sources, data availability, and data frequency are being introduced for greater clarity and preciseness.

Government regulations have had a significant impact on the evolution of AVMs as a result of the 2008 housing crisis and in an effort to add greater transparency to mortgage transactions. The Interagency Appraisal and Evaluation Guidelines (Interagency Guidelines) were revised and distributed in December 2010 by the Federal Financial Institutions Examination Council (FFIEC) to provide more detailed information on how agencies expect lenders to deal with appraisals, broker price opinions (BPO) and AVMs. In terms of AVMs, the Interagency Guidelines provide perspective on model validations and when an AVM can be used as an evaluation tool.

Additionally, the Dodd-Frank Wall Street Reform and Consumer Protection Act requires AVMs to integrate quality control standards for a higher level of credible AVM estimates, guard against data manipulation and conflicting points of view, as well as require sample testing. As a result, both the focus on AVMs and their use is evolving before our eyes.

Us: What are the key factors that a firm like Veros uses to make a computer-generated valuation of my house?

William: There are two basic elements to any computer generated valuation, the model itself and the data that feeds the model. Veros’ AVM models are used in a broad range of proprietary algorithms to develop an estimate of value for a given property. Veros has developed a variety of individual models including statistical models, time-based models and comparable sale models, among others. Final values are the result of a unique and proprietary process of weighting and optimization that is designed to produce the highest confidence in the value produced. Veros’ databases include publicly available data from property assessors, as well as proprietary data unique to Veros.

Us: How do you account for the quirk or qualities that aren’t obvious?

William: I believe this question is directed to the additions and improvements that are unique to a given home. For example, if I renovate a 1950’s home in a neighborhood that lacks modernized homes, how do we account for this in the AVM? There are several limitations to computer generated values and this is one of the more challenging of them. However, this works in both directions. For example, just as a nicely improved older home in a neighborhood where the sale transactions are for homes without renovation results in an AVM value that is probably below the true value; a neglected home in a neighborhood where the sales reflect well-maintained homes will get a higher AVM value than the true value. In real estate valuation terms, this is the principle of progression and regression. Basically, this means that lesser quality homes will benefit by association with better quality homes (progression) and higher quality homes will suffer by association with lower quality homes (regression). The principle of conformity in real estate says that values will be the most stable in a neighborhood where the homes complement each other in terms of size, style, age and condition. AVMs produce the most reliable values in neighborhoods where there is reasonable conformity within the neighborhood.

Us: Lots of experts got the last downturn wrong. What did Veros learn from the slump and how is that incorporated into the current algorithms?

William: Among many things, we learned to continue trusting our models. Many of our competitors were quick to adjust their models to conform to the rapid changes in home prices. As noted in Veros’ 2010 report on Residential Real Estate Forecasting Using VeroFORECAST, while some of our raw numbers proved incorrect e.g. our April 2007 forecast for Miami was for a -6.0% change but that market actually changed by -17%. We were still the only forecast that correctly saw the price decline ahead of its occurrence. Our forecast has shown consistent strength and accuracy in rapidly rising markets, markets that have flattened, and markets that have rapidly declined.

Us: What are your computers tell you about the trend line for housing prices?

William: Veros real estate market forecast product, shows that once again Bismarck, North Dakota leads the housing market for strongest home appreciation, followed by Honolulu, Hawaii and Fargo, North Dakota. These areas of North Dakota are showing strength due to a notably low unemployment rate at 3.5 percent, a growing economy and population fueled by an oil boom and strong demand for agriculture. Honolulu attributes its strength to an unemployment rate of 5.7 percent, its lowest housing supply in the past five years and a high rate of affordability. Nationally, the Veros Future Home Price Index continues to show stability in its quarterly results. With no significant drags on HPI and few markets showing strong appreciation, a very slow recovery for housing prices is anticipated. Unemployment and housing supply continue to be key discriminators between the top and bottom 10 housing markets. VeroFORECAST indicates a maximum of five percent appreciation for growing housing markets and five to six percent decline for weaker markets. The most strength in housing markets can be found in the Great Plains, including North and South Dakota, Texas, Wyoming, Nebraska, Louisiana and Iowa. The nations weakest housing markets are being found in Nevada, inland areas of California, Washington and Oregon, accounting for seven out of the bottom 10 housing markets.