About Automated Valuation Models (AVMs)
When surveyors and estate agents assess the value of a property, they assemble the information they have about the property, such as comparable recent local sales, and apply a set of guiding principles to this information to come up with a prediction of the price that property would fetch in the current market. Automated valuation models enable computers to perform the same task of assembling relevant information, applying rules to the information and producing an estimate.
Computers can accomplish this task in several ways. One method treats the total value of a property as a function of its specific characteristics. The number of rooms, overall size, and presence of features like swimming pools, garages, views and so on can influence the value of a property. This method is good at customizing a valuation to the specifics of a given property, but it depends on complete and accurate information for the property being valued and all the properties used to calibrate the model to begin with.
Another method assumes that past sales values of the property in question are fair representations of the property's value at that point in time, and, as long as the property has not changed substantially since that sale, the current value depends only on the prior value and the change in property values since the time of that sale. For example, if a property last sold for £100,000 and values have since risen 25%, the property is likely currently worth £125,000. This method is good at using the best information available about the market value of a property (the prior selling price), but is less accurate if a property has had major improvements made or if the prior sale happened under non-standard market conditions, and was not reflective of the market value at that time.
Yet another method automatically selects a group of similar properties with recent sales, and uses this group to impute a value for the target property. This method also attempts to take into consideration the specific characteristics of the property in question and the current state of the market. Missing or inaccurate information about the characteristics of either the property to be evaluated or potential comparable properties will impair this method's accuracy.
While every automated valuation method has strengths and weaknesses, there are many reasons automated valuations are valuable and worthwhile:
- AVMs are objective. They have no motive to over or undervalue property. They consider only the facts available to them about the property, and are unbiased by their relationship to the property. An owner looking to sell, a potential buyer, a mortgage broker, a tax assessor all have motives that may colour how they perceive the value of a property, even if not consciously.
- AVMs can consider a lot of data. Computers can think through the millions of pieces of relevant data much more effectively than we can. People tend to make up for this difference by instead focusing on just a few, highly relevant pieces of information, but there is some information loss as human beings are likely to disregard data that's relevant, even if not central, to the decision at hand.
- AVMs are fast. It can take a professional several days to assemble relevant information, assimilate it, and determine an assessment of value. This same process takes an AVM several seconds.