2018 has been a year of back flipping robots and electronic assistants managing our personal and professional lives. Furthermore, a meeting is no longer a meeting unless someone mentions drones, AI or machine learning.

All this caused by unprecedented access to data, and the computing power to drive insights is raising expectations to new levels for automation, accuracy and hyper-personalisation.

At CoreLogic we are extremely fortunate to be the custodians of one of richest data assets in Australia and on one of the most talked about subjects, property.

When it comes to property valuations, this places us squarely in the middle of a fascinating dynamic of data expectation, customer emotion and a highly active regulator.

The expectation is to value all property with a high degree of accuracy given the myriad data and advanced modelling capability. We know factors such as sales history, flight paths, proximity to water, aspects, power lines, construction, hazard, market conditions and therefore, should be able to predict everything that impacts the value of a property.

The question however is; “for what purpose are you valuing a property?” 

  • I want to buy - it’s too expensive
  • I want to sell - you’re undervaluing
  • I want to satisfy the regulator for a loan origination – how accurate and governed is your model?

In some way everyone is correct.

In a large post GFC regulation change in the US, the two government sponsored enterprises (GSE) Freddie Mac and Fannie Mae have opened up the use of automated valuation models for certain loan types. One of the major driving forces is the diminishing number of qualified valuers to take on all of the work, but another is their access to huge amounts of contemporary, accurate property data.

Fannie Mae also loudly vocalise on the requirement to continually capture new and up-to-date property data in order to ensure their valuation models continue to perform to a level that meets their securitisation standards.

Globally, the credit bureaus have done an incredible job leveraging the available consumer data to create predictive scores and behavioural insights. With the dawn of open banking, understanding consumer transactional behaviour will open up further possibilities, particularly for thin credit file consumers, to gain access to automated decisions and offers.

Fundamentally, the differences between property and people are the amount of similarities that exist within each group and the consistency of predictive data for each group. Demographic segmentation has been around for a long time, likewise so has predictive modelling for credit behaviour. Properties in Australia vary from the highly homogeneous to incredibly unique and can be located within a matter of a few hundred metres.

Some things to think about:

  • If you are displaying valuation modelled estimates to consumers, brokers and lenders, ensure they are accompanied with the right measure of accuracy and some education tools on their predictive power
  • If you are relying on valuation modelled estimates in loan origination, be sure you have appropriate usage rules and governance frameworks in place for model development and performance monitoring
  • Ensure you are capturing and managing your property data to empower existing and future data driven decision making

Our vision at CoreLogic is to power housing through data connectivity and analytics. 

Modelled estimates are extremely powerful if used appropriately on data rich properties in the correct geographies. CoreLogic has some incredible tools to understand property data coverage and the predictive power for property valuation of this data. Some of the connectivity tools are also blurring the lines between the contributions to valuation of the humans and the machines

In the near term, I am aligned to the view of the GSE’s in the US on accurate and timely data, full transparency on coverage and appropriate measurement of the predictive power of the data for the purpose of valuation.

So for the time being, the humans are still running the show, highly empowered by the machines. However the rate of change in data, regulation and technology is shifting this balance on a daily basis.