Automated Valuation Model (AVM) in Real Estate Sector

This article discusses the role and potential of Automated Valuation Model (AVM) in the real estate sector. It explains how AVM uses mathematical modeling and data analysis to estimate the market value of properties. The advantages of AVM over traditional valuation methods are also highlighted.

Automated Valuation Model 1
AUTOMATED VALUATION MODEL (AVM)
Name
Course
Professor
University
City/state
Date

Automated Valuation Model 2
Automated valuation model (AVM)
Role of real estate valuation modelling techniques
Real estate has significantly been transformed over the past few decades and is largely being
driven by technology today. There are numerous technologies that have been developed to
improve the accuracy and reduce turnaround time of property valuations. Some of these
technologies include: use of robotics (drones), artificial intelligence, actual reality (AR), virtual
reality (VR) and internet of things (IoT) to conduct property inspections (Garg, 2018), use of
data and analytics for property valuations, and use of cloud computing to generate market prices
of properties (Wall, 2018). Just as many people are using the internet and mobile devices to look
for homes to buy nowadays, appraisers are also using the internet and other technological tools to
determine the value of homes and other real estate properties. The main focus of this discussion
is on the application of automated valuation model (AVM) and its potential in the real estate
sector. AVM is a technological tool or technique that uses a combination of databases and
transactions of an existing and similar real estate properties and mathematical modelling to
estimate the market value of a real estate property. This is a more sophisticated technique of
valuing real estate properties. There are different types of AVMs and most of them determine the
value of a real estate property by comparing values of similar properties in the neighborhood of
the subject property. Sales comparison approach is a very useful property valuation technique
because it shows evidence of the current market value.
Real estate valuation modelling techniques are used for sale and purchase, financing,
investment, expropriation, tax assessment, transfer, estate settlement or inheritance of properties.
The main role of real estate valuation modelling techniques is to generate the market price of a
property at a particular time (Bulut, et al., 2011). These property valuation modelling techniques

Automated Valuation Model 3
have been designed to factor in a variety of factors that affect the price of the subject property.
AVM is one of the real estate valuation modelling techniques. In this technique, the price
estimate is generated by weighing and analyzing a repeat sales index and using a hedonic model
(this is a type of multiple or statistical regression analysis). Hedonic modelling systematically
analyzes the impact of each attribute of the property on its market price or value (Antipov &
Pokryshevskaya, 2012). Some of these attributes include: location (accessibility, site,
neighborhood and distance to public transport and central business district), physical (size of
property or individual rooms, number of bedrooms, number of bathrooms, number of rooms,
year built, garage, swimming pool, parking space, type of HVAC (heating, ventilation and air-
conditioning), lawn, gardens, courtyards and quality of construction), time (market conditions
and time of sale), and tenure (leasehold or freehold). Hedonic model uses a combination of
artificial intelligence techniques, artificial neural networks and regression analysis to generate a
price model of the subject real estate (McCluskey, et al., 2012). Most sophisticated AVMs being
used today employ models based on data mining and machine learning techniques (Matysiak,
2017).
In simpler terms, AVM is a computer program or computerized system that uses
databases of existing real estate and mathematical modelling to create property valuations
(Glumac & Rosiers, 2018). The data can be obtained from pubic repositories like land registries
or subscription/proprietary sources (Williams, 2018). The program estimates the price of the
subject property very quickly using the factors that are known to influence the real estate value
the most. This eliminates or reduces the time and cost of an appraiser going to the property
physically to inspect the home or house, go through the appraisal checklist, evaluate every item
and estimate the price of the property. However, it is important to note that AVMs are not as

Automated Valuation Model 4
thorough as traditional property valuation methods but they can generate estimates of properties
within 10% of the actual sale price.
Typically, an AVM comprises of the following: indicative market value for numerous
properties, information about the subject real estate including its recent sales history, indication
value of the tax assessor, and comparable sales analysis of similar real estate. AVM is not new
but has been in use for more than 20 years now and its use is prevalent in most developed
countries such as the U.S., UK, New Zealand, Australia, China, Japan, Sweden, Canada,
Germany, Denmark, South Africa, South Korea and Singapore (Bidanset, 2014), and is also
gaining ground in developing countries. Some of the known providers of AVMs include: JLL – a
global property services company based in Australia (JLL, 2018); ATTOM Data Solutions,
Zillow, Trulia, CoreLogic, Realtor.com, HouseCanary and Homesnap based in the U.S. (Adams,
2019); (ATTOM Data Solutions, 2019); and Clear Capital AVM, Hometrack, Veros Real Estate
Solutions, Black Knight, Rightmove and AVM Analytics in the UK (Pohlmann, 2018), among
others. Adoption of AVM is also gradually increasing and shaping Singapore’s real estate sector
and may soon become mainstream in the country’s property valuation space.
It is also worth noting that the role and use of AVM is not limited to the private sector.
Governments also use or can use AVM to value and rate properties regularly for the purpose of
recalculating taxes. This will not only help the government to collect more revenue but also
enable it make better planning decisions (Novak, 2017).
Advantages of AVM over traditional valuation methods
The main goal of property valuation process is to calculate the most accurate value of the
property at any given time using the smallest amount resources possible. Most AVMs are

Automated Valuation Model 5
considered to be better than traditional valuation methods. The undisputable fact is that both the
AVMs and traditional valuation methods have advantages and disadvantages. Examples of
traditional property valuation techniques are comparable technique, profit technique,
investment/income technique, cost or contractor’s technique, residual or development technique
stepwise regression technique and multiple regression technique. On the other hand, advanced
property valuation techniques include: hedonic ricing technique, fuzzy logic, artificial neural
networks (ANNs), spatial analysis techniques, and autoregressive integrated moving average
(ARIMA) technique (Yeh & Hsu, 2018). AVM is definitely an advanced property valuation
technique. There are numerous advantages of AVMs over traditional valuations methods. Some
of the main advantages of AVM include the following:
Saves resources, time and money: the amount of resources, time and money used to
perform AVM appraisals is very low compared to that of traditional valuation methods (Mooya,
2011). This valuation technique eliminates the need of an appraiser to hire property inspection
equipment, travel to the where the property is located and conduct physical inspection of the
property, and all the costs, time and resources associated with these activities. When using an
AVM, the estimate value of a property can be generated in seconds (reduced turnaround time)
and at a very lower cost relative to the cost of traditional appraisals.
Can be used by anyone: using AVM does not require prior knowledge on how the AVMs
are designed or how they work. The user only needs to know the attributes of the property to be
valued. For example, the AVM system (called X-Value) provided by SRX Property can estimate
the value of a property in Singapore by factoring in the following attributes of the subject
property: name and address of the property, type of property, number of units, number of floors,
gross floor area (in m2), and outdoor area (SRX Property, 2019). This means that anybody who

Automated Valuation Model 6
wants to estimate the value of a property in Singapore can use this system anywhere anytime as
long as he/she has the aforementioned details of the subject property. For traditional valuation
methods, only trained and certified property valuers or appraisers can perform the valuation.
No human bias and partiality: AVM is a hedonic price technique of real estate valuation
and therefore it does not contain the human element that is very common in traditional property
valuation methods but relies on computer automation. Sometimes human appraisers using
traditional valuation methods can manipulate data for favoritism but this does not happen with
AVM. The AVM does not have any commercial r vested interest in the property being valued
and completes the valuation based on facts and not self-interest or emotions (Lee, 2016).
Versatility: AVM technology is versatile and can be used by appraisers, real estate
agents, financial or lending institutions, mortgage lenders, brokers and investment experts for
valuing a wide range of properties including commercial, residential, hotels, industrial and
development sites. For the case of traditional valuation methods, most valuers are specialized in
properties they appraise and are restricted to operate within specific geographical locations.
AVM is also flexible such that its application is not hindered by the geographical location of the
user.
Improves valuation accuracy: there are several ways in which AVM increases the
accuracy of estimating the value of a property. There are multiple ways in which AVM improves
the accuracy of property valuation. One of these ways is that AM narrows down valuation to
specific type of properties with particular attributes. For example, it can narrow down the
valuation to detached houses with one floor of 150m2 (gross floor area), three bedrooms, two
bathrooms, a garage, swimming pool and a courtyard. This means that during valuation, only
properties with these attributes within a specific area will be considered to estimate the

Automated Valuation Model 7
comparable value of the subject property. Another approach is that AVM uses several valuation
models to select the most suitable and accurate modeling technique for the geographic area
where the property being appraised is located. Last but not least, AVM can make comparison of
as many as 100 similar properties thus generating a more accurate market price estimate of the
subject property. All these increases the accuracy of estimating the value of the property.
Latest sales data: AVM uses data of recent property sales in the region where the subject
property is located. The properties used in the comparison can be up to hundreds in numbers.
The properties may have been sold or bought in the past few days, weeks and months. This helps
in providing more accurate estimate of the property value. However, traditional valuation
methods can only use very few properties for comparison purposes because the number of
properties used also affects the time and cost of valuation.
Valuation changes: with AVM, real estate agents and homeowners are able to track
changes in the value of the property thus being able to know the right time to sell it (Taylor,
2018). As stated before, AVM can value a property in seconds. This means that a real estate
agent or homeowner can use AVM to estimate the value of the property and compare it with the
price being offered by potential buyers at the time. It then becomes very easy to decide when to
sell the property. Likewise, buyers can also use AVM to determine the most appropriate time to
buy a property within a certain location. This may not be possible when using traditional
variation methods because they take more time to value a property and the cost of valuation is
higher than that of AVM.
Transparency: AVM is also more transparent than the traditional valuation methods that
are done without revealing all the information use to estimate the property’s value. In AVM, all
the methodology adjustments and comparable transactions used in the valuation are well

Automated Valuation Model 8
documented. More AVM vendors are also coming out to explain how their modes work, which
enhances understanding and scrutiny by users.
As stated before, AVM also has some disadvantages and it is important to highlight some
of them. One of the main drawbacks of AVM is that it does not consider the current condition of
the property when estimating its value. This is because AVM does not include a physical
inspection of the property to determine its condition but assumes that the condition of the
property is average. This is a major drawback of AVM because a property’s current condition is
a major factor that affects its value. It is mandatory to establish and consider the current
condition of a property when estimating the value of the property because it determines whether
the property has been refurbished recently or if it will require renovation soon, which has
significant effect on the market price of the property.
Another disadvantage of AVM is that there are chances of using inaccurate data in the
property valuation process. AVM works by integrating big data stored in cloud computers. Some
of these sources may be manipulated intentionally because of concealed incentives provided by
some stakeholders so as to influence recorded sales prices. The data is also susceptible to illegal
access (cyber-attacks), which will provide inaccurate data and valuation.
Implications of using AVM
Accuracy and reliability remain the major issues of concern in relation to the use of AVM in
valuing properties. Some professional bodies in Singapore have raised concerns about the
accuracy and reliability of AVMs. In 2016, the Singapore Institute of Surveyors and Valuers
(SISV) released a statement stating that they did not recognize valuations generated by computer
programs because they were not in accordance with practice guidelines and valuation standards

Automated Valuation Model 9
of SISV (Khoo, 2016). This came after SRX Property launched a property valuation service
provided by AVM. Just like any other property valuation technique, AVMs are not 100%
accurate and the accuracy largely depends on the type of data used by the providers to create the
AVM. The main concern, which is also major drawback of AVM, is that the AVM does not
factor in the current condition of the property and instead assumes ‘average condition’ scenario.
Most studies have shown that the valuation accuracy of AVM depends on several factors
including the type of property, location, attributes of the property, source of data used, and the
valuation criteria of the specific AVM. Nevertheless, AVM vendors have continued to develop
AVMs with high valuation accuracy. For instance, most of the AVMs used in the U.S. and UK
have recorded valuation accuracies of± 10 %. Therefore AVMs does not provide the best
valuation estimates but the estimated they generate are within acceptable accuracy limits. What
is important to note is that AVMs are estimates hence margins of error in valuations are
inevitable.
Situations not suitable to use AVM
There are also cases where AVM is not suitable for use in property valuations. One of such cases
is when there is no adequate or any data available of similar properties. This can happen when
the property being valued is unusual and is not similar to any other property in the neighborhood.
For example, when the subject property is a four storey building with eight bedrooms yet all
other buildings sold/bought recently in the neighborhood are one or two storey with three to four
bedrooms, then AVM is not suitable for use in valuing the subject property.
AVM should also not be used when its provider cannot access adequate, complete and
high quality off-market data. One of the factors that reduce the accuracy of AVM is reliance on

Automated Valuation Model 10
public record sources, which are usually incomplete and inaccurate. These records react to the
changing real estate market trends very slows and therefore do not provide accurate or latest data
about property prices in an area. Therefore AVM should only be used when its provider can have
access to off-market data from both public and private property firms.
Additionally, AVM is not suitable for use in an area dominated with new constructions.
These include upcoming suburbs in major cities. Such areas do not have adequate property sales
data needed by AVM to estimate the value of the subject property.
The future of AVMs
This is the age of automation and AVM has great potential of disrupting the property valuations
space in Singapore and become an accurate and reliable valuation tool. Moving forward, one of
the ways to improve adoption of AVM in Singapore is to formulate proper legislations and
regulatory framework. The government of Singapore can promote use and acceptance of AVM
by formulating policies that specify the standards and practice guidelines of AVM. This should
be done in collaboration and consultation of relevant professional bodies such as SISV. This kind
of a move will also increase awareness and acceptance of AVM. The accuracy and reliability of
AVM can also be increased by factoring in the current condition of the property (by using
robotics, VR or AR to perform physical inspection of the property) and integrating human or
professional judgment. As most activities and transactions turn digital, there is great potential of
AVM becoming mainstream in real estate valuations space in the near future.

Automated Valuation Model 11
References
Adams, B., 2019. What is a Real Estate AVM? A List of the Major AVM Companies. [Online]
Available at: https://hooquest.com/real-estate-avm/
[Accessed 28 May 2019].
Antipov, E. & Pokryshevskaya, E., 2012. Mass appraisal of residential apartments: An application of
random forest for valuation and a CART-based approach for model diagnostics. Expert Systems with
Applications, 39(2), pp. 1772-1778.
ATTOM Data Solutions, 2019. Attomized AVM. [Online]
Available at: https://www.attomdata.com/data/analytics-derived-data/avm-property-valuations/
[Accessed 28 May 2019].
Bidanset, P., 2014. Moving automated valuation models out of the box: the global geography of AVMs.
Belfast, Northern Ireland, Ulster University.
Bulut, B., Allahverdi, N., Kahramanli, H. & Yalpir, S., 2011. A Residential Real-Estate Valuation Model with
Reduced Attributes. International Journal of Mathematical Models and Methods in Applied Sciences,
5(3), pp. 586-593.
Garg, A., 2018. New Real Estate Technology: Disruptive Ideas Transforming the Industry. [Online]
Available at: https://www.entrepreneur.com/article/320824
[Accessed 27 May 2019].
Glumac, B. & Rosiers, F., 2018. Real Estate and Land Property Automated Valuation Systems: A
Taxonomy and Conceptual Model. SSRN Electronic Journal, 1(1), pp. 1-21.
JLL, 2018. JLL launches new proptech valuation model for residential in Australia. [Online]
Available at: https://www.jll.com.au/en/newsroom/jll-launches-new-proptech-valuation-model-for-
residential-in-aus
[Accessed 28 May 2019].
Khoo, L., 2016. SISV rejects computer-generated valuations. [Online]
Available at: https://www.businesstimes.com.sg/real-estate/sisv-rejects-computer-generated-valuations
[Accessed 28 May 2019].
Lee, J., 2016. Smart property for a smart nation. [Online]
Available at: https://www.todayonline.com/business/smart-property-smart-nation
[Accessed 28 May 2019].
Matysiak, G., 2017. Automated Valuation Models (AVMs): a brave new world?. Kracow, Poland, Cracow
University of Economics.
McCluskey, W., Davis, P., Haran, M., McCord, M. & McIlhatton, D., 2012. The potential of artificial neural
networks in mass appraisal: the case revisited. Journal of Financial Management of Property and
Construction, 17(13), pp. 274-292.
Mooya, M., 2011. Of Mice and Men: Automated Valuation Models and the Valuation Profession. Urban
Studies, 48(11), pp. 2265-2281.

Automated Valuation Model 12
Novak, M., 2017. From Manual Appraisals to Automated Valuation Models (AVMs). [Online]
Available at: https://medium.com/geophy-hq/from-manual-appraisals-to-automated-valuation-models-
avms-4ec2c0b2720e
[Accessed 28 May 2019].
Pohlmann, J., 2018. How Property AVMs Differ in the Real Estate Industry. [Online]
Available at: https://www.attomdata.com/news/most-recent/property-avm-providers-comparison/
[Accessed 28 May 2019].
SRX Property, 2019. X-Value pricing. [Online]
Available at: https://www.srx.com.sg/xvalue-pricing
[Accessed 28 May 2019].
Taylor, M., 2018. What Is AVM (Automated Valuation Model) in Real Estate?. [Online]
Available at: https://realtyna.com/blog/what-avm-real-estate/
[Accessed 28 May 2019].
Wall, K., 2018. How data and technology innovation are transforming valuations. [Online]
Available at: https://www.housingwire.com/blogs/1-rewired/post/43319-how-data-and-technology-
innovation-are-transforming-valuations
[Accessed 27 May 2019].
Williams, S., 2018. Automated Valuation Models: Opportunity or obstacle?. [Online]
Available at: https://www.rics.org/asean/news-insight/latest-news/news-opinion/avms-the-demise-of-
the-valuer/
[Accessed 28 May 2019].
Yeh, I. & Hsu, T., 2018. Building real estate valuation models with comparative approach through case-
based reasoning. Applied Soft Computing, 65(1), pp. 1-44.