After analysing several automated property valuation websites, Consumer Group Choice has raised questions about their accuracy.

Although e-valuations can provide useful information and guidelines, Choice found that different websites can give a wide range of values on the same property and can differ significantly when compared to licensed valuations and actual sale prices.

“Without third-party verification of their price estimates, or a comparison against real valuations and sale prices, choosing a website is a bit of a lottery,” Choice said. “So use the websites to gather information and aid your research, but don’t expect a number you can rely on.”

In researching the websites offering automated valuations, Choice looked at property reports bought by a couple preparing to bid for a house in Sydney, and compared them with an independent valuation.

They found that the online estimates obtained for the same three-bedroom house differed by over $110,000 — ranging from $548,000 to $662,661. The least accurate e-valuation was more than $140,000 off the eventual sale price.

Some of the websites evaluated gave extremely wide value ranges for the same property; for example, $543,000 to $782,000, and $484,600 to $706,000.

The licensed valuer’s estimate for the property was $630,000. It sold for $690,000 two weeks later.

Peter Rossini, Program Director of the University of South Australia’s School of Commerce and author of several studies on automated property valuation models, is quoted in the review.

“While Automated Valuation Models (AVMs) may be useful for lending institutions in a portfolio or mortgage situation, I have grave reservations about their general use by consumers,” Rossini said.

Choice notes that all the competing websites get most of their data from state government property sales registers which is supplemented with information about recent sales from estate agent contacts, and details about property attributes. Statistical modelling is used to generate estimates of value.

But Choice said there are some things that a statistical model can’t do. They don’t take the condition of the property into account — there could have been renovations and improvements, or the property could be in disrepair. The automated systems remove human bias and subjectivity, but emotion and personal bias often affect the selling price. Another disadvantage of automated valuations is that the figures are slower to adjust to oversupply or undersupply of properties in a particular suburb. “These supply and demand factors also affect prices,” Choice said.

Lending Central