Computer Generated Appraisals

VCL’s data analyst maintains the algorithms and analysis the national property sales data. The results from this analysis are used to produce HomeValuer, a leading New Zealand Automated Valuation Model (AVM). HomeValuer may be purchased online through CoreLogic or Zoodle.

HomeValuer is an Automated Valuation Model more commonly referred to as an AVM. An AVM is a computer program designed to simulate the traditional valuation process and provide instant residential property assessments.

HomeValuer is a product developed by Real Estate Rarket Intelligence (an associated company of VCL). VCL maintain the algorithms and analyses the raw sales data on a weekly basis. The analysed data are then passed electronically to Terralink International who produces the final reports.

The HomeValuer report instantly provides you with the following:

• Computer generated value assessment.
• A confidence score depending on the quantity and quality of sales data found.
• Physical and legal description of the property.
• Sales history relating to the subject property.
• Current rating valuation.
• Local market details.
• Sales used to calculate the value.
• Location map of comparable sales.
• Legal details relating to the property.
• Arial imagery of the property.
• Selected census information.

If you would like to view a sample HomeValuer report before purchasing follow this link:
http://www.terranet.co.nz/terranet3/images/samples/hvmfull.pdf?1349397384

If you would like to purchase a HomeValuer report as it relates to your property follow this link: http://www.terranet.co.nz/terranet3/products/automated_property_valuations

If you would like more information regarding AVMs, here are answers to some frequently asked questions.

How accurate are AVMs generally?

AVMs can vary widely in accuracy, reliability and usefulness based upon numerous factors including data quality, real estate fundamentals and complexity of the underlying technology.

How accurate is HomeValuer and how do we know?

We test the model against the new sales available and measure how well we predicted these sale prices. As a generalisation the expected accuracy level in larger population centres is better than in areas where sales are less frequent. The accuracy percentage for each area is given within the confidence score when you purchase your AVM.

What are the limitations of all AVMs?

The values are totally software generated and the property is not physically inspected. The results so derived have an underlying assumption that the dwelling is in a similar average condition (Ceteris paribus) as to that of the comparable sales referred to in the report.

There are four possible areas of error. These can be categorised as follows:

• The property record might be incorrect. One or more of the descriptive variables might have been incorrectly recorded to the data base. For example a dwelling might have an actual floor area of 200 square metres but if this was recorded as being 100 square metres in size, the whole record would be invalid. As we only analyse the data passed to us the value derived by the AVM will clearly be incorrect. Or in IT language “garbage-in equals garbage-out”.
• The physical condition of a property might be worse than that of the competition. For example a structural element might be in need of replacement (e.g. roof, exterior cladding, plumbing, etc.), or the property might simply be in poor decorative presentation. As it has not been inspected such adjustments cannot be made.
• Surrounding or neighbouring properties may be of a nature that has a negative effect on the subject.
• The property may have super adequate features. Super adequacy is defined as being “a superior quality feature or an additional feature not possessed by its competitors”. It would be expected that such a property might achieve a premium. Again, as the property has not been inspected, these adjustments cannot be made.
• Subjective assessments not possible. Property valuation is sometimes described as more art than science. Fine value adjustments are often subjective in nature. A specific property might achieve a premium because it ‘feels’ better or is better presented than the comparisons. AVMs are unable to make such discrimination.

Why bother to use AVMs at all?

The major benefits of AVMs are as follows:

• Cost – a significant barrier to obtaining independent advice when undertaking due diligence either as an intending purchaser or seller is the cost of that advice. For example, a purchaser who is considering multiple properties would be unlikely to commission a valuation on each at a cost of $450-$500 + per property. However at a cost of around $50 per property, the option becomes affordable.
• Speed – results for all intents and purposes are instant. Opportunities can sometimes be lost through time taken on research, particularly in a competitive situation. A quote attributed to Sun He Yuan Zhu “Victory goes to the well-prepared”. The AVM provides you with a great deal of useful and relevant information (over and above the value estimate). You should use that information to undertake your own research and in deriving your own estimate of value.
• Independence – results can sometimes be skewed by one party to a transaction exercising undue influence on a human expert. This is not possible with an AVM. Indeed, neither VCL or CoreLogic are able to intercept and modify the results from any individual transaction.

Is HomeValuer suitable for Commercial or Rural property valuations?

The simple answer is NO. AVMs require a reasonable amount of homogenous data to operate successfully. Commercial and Rural properties often have characteristics which are unique to individual properties. Therefore it is not feasible, at this time, to automate valuations for these categories of properties and still remain within consistent margins of accuracy.

Some Tips on getting the best out of AVMs

AVMs give you access to a rich base of knowledge and in negotiation ‘knowledge is king”.

• One of the most effective ways to use Automated Valuation Models (AVMs) is to supplement traditional appraisal and valuation methods. If the value indicated by the AVM is at odds with a Real Estate Agents appraisal or an asking price, it might indicate that you should seek a second professional opinion.
• AVMs will usually provide you with a list of comparable sales (if there are any). Drive past these sales yourself to confirm the relative quality of these as they compare with the subject. You can often tell a great deal from a road side inspection. The AVM will provide you with additional data such as house size and age, section size and rating value.
• Use the AVM in conjunction with your own physical inspection. For example, if on inspection it is apparent that a roof needs to be replaced, an AVM supplemented by a builders estimate to replace the roof will get you closer to the real value of the property.

Warning: Be aware that HomeValuer uses the best technology available but it is not infallible. The application of common sense can minimise any risks.

How do AVMs work?

There are two possible mathematical models and two statistical models which tend to be used by competing products. Each model has inherent strengths and weaknesses.

The mathematical models are as follows:

• Indexing – this simply tracks price movements within a locality and applies this to a constant (usually Rating Value or most recent sale price) which it uses to anticipate the most likely selling price of the subject. (e.g. Rating Value x (1+% movement) = Value).
• Expert Systems – these are also known as heuristic models. These emulate the processes of a human expert. Comparable sales are selected based on similarities of physical characteristics (e.g. size, age, condition of the property, etc.) The results are mathematically adjusted for differences and a value is derived.

The statistical models are as follows:

• Regression Analysis – Utilise complex mathematical equations to define the statistical relationship between one or more property characteristic and the value of the property. As a point of interest regression models are used for your Rating Value Assessments.
• Neural Networks – Are designed to identify the statistical relationships between house prices and their associated physical characteristics. Whereas regression models are described as being static, neural networks (sometimes described as non-linear adaptive models) are dynamic.

Which Model does HomeValuer use?

HomeValuer may be described as a ‘Cascade Model’ which uses a hierarchy of models. It integrates Indexing, Expert and two different Non-linear Adaptive models onto a single platform. It analyses the results from the four models and selects the model which gives the results with the most consistent accuracy from within the immediate locality. This tends to give a more consistent level of accuracy than that produced by any single model if used alone.

AVMs are not meant to replace and should NEVER be considered to be a valuation. They have their place but should be use with caution. All AVMs associated with Real Estate market Intelligence and VCL are at all times provided subject to the attached terms and conditions.

CLICK TERMS AND CONDITIONS http://www.terranet.co.nz/terranet3/terms