How Big Data algorithms can bolster Singapore's mortgage system

BIG Data algorithms give Singapore a unique opportunity to strengthen its home loans financing system.

Let me explain. In June 2013, the Monetary Authority of Singapore (MAS) introduced the total debt servicing ratio (TDSR) framework to strengthen the credit underwriting practices of financial institutions, encourage financial prudence among borrowers, and help cool the property market.

In 2012, an MAS review had concluded that the major banks had "adequate" policies for assessing the creditworthiness and ability of borrowers to repay. But the methodologies and practices of the banks varied to such a degree that MAS moved quickly to strengthen and standardise their credit underwriting practices by implementing the TDSR. All of this was done against a backdrop in which home loans, as a percentage of GDP, increased from 34 per cent in 2009 to 46 per cent in 2013.

TDSR has been effective in cooling property prices and standardising lending rules with more stringent qualification requirements. But it is an imperfect remedy. It has not standardised valuations: a key part of the underwriting process. Furthermore, in effectively cooling the real estate market, TDSR and other cooling measures have increased the risk of mortgage defaults.

As interest rates rise, mortgage payments go up. At the same time, homeowners see the market value of their homes declining as a direct consequence of price drops. History tells us that this perfect storm of rising mortgage costs and devalued collateral (that is, home values) will lead to an increase in non-performing loans. Too many non-performing loans can wreak havoc on Singapore's financial system. If not managed properly, this perfect storm could potentially threaten the economy as much as a bubble in the property market would.

The key to managing the Singapore real estate market would be the successful implementation of a transparent pricing mechanism. Up to now, the real estate market has been inefficient and devoid of a universal pricing standard. As a result - much to the detriment of the overall market - buyers and sellers have made pricing decisions based on non-standardised, incomplete and (often) irrelevant information.

Furthermore, there has been no market-pricing mechanism against which policymakers could stress-test the impact of changes in supply, demand, policies and external shocks on the market. Valuations used in credit underwriting have also been inconsistent - resulting too often in the matching of valuations with the asking price.

While it has been relatively straightforward to standardise the qualification requirements of borrowers and the structure of home loans under the TDSR, no regulation, rule nor training can "standardise" the human valuation of property collateral. A valuer does not have the ability to organise the entire market, and can be swayed by human error and bias. The reason is that each valuer operates as an independent actor, relying on single sources of raw data that are often inadequately cleaned and suffer in performance because they are not integrated with other sources of Big Data or subject to sufficient checks and balances.

Only neutral algorithms and workflow applications - like that of SRX Property's X-Value - can measure and reflect the interactions of hundreds of thousands of decisions taken by consumers, agents and other market participants.

A computer-driven price mechanism can instantaneously mash up data from over 30 Big Data sources; bring in important geospatial considerations; make adjustments for critical variables like location, size, floor and age; and generate in seconds a single value with transparent supporting documentation. Computer algorithms are much better geared to employ objective methodologies and ensure results with integrity.

As a result, this computer power opens up new realms of possibilities. Valuers can use tools such as X-Value as a second opinion in rendering assessments. Mortgage underwriters can use a standardised approach to valuations in addition to the credit and loan structures mandated by TDSR. Risk managers at banks can mark-to-market the collateral in their mortgage portfolios in real time and receive computer-calculated early warning indicators of potential troubled loans.

Policymakers can use real-time market valuations and the banks' marked-to-market portfolios to stress-test the impact on the financial system of changing interest rates, supply, non-performing loans, and default rates. The end result would be more certainty and stronger forecasting capabilities.

Finally, the ubiquitous dissemination of a pricing mechanism would provide transparency to all real estate participants and restore confidence in the market.

Every market needs a standard, transparent pricing mechanism. Much like 220 volts is Singapore's electrical standard, X-Value should be our real estate pricing standard. The more we embrace a transparent pricing mechanism, the more stable our home mortgage and financial system will be.