Getting Deep and Dirty On Mortgage Risk

We have been busy adding in new functionality to our Core Market Model, which is our proprietary tool, drawing data from our surveys and other public and private data sources to model and analyse household finances.

We measure mortgage stress on a cash flow basis – the October data will be out next week – and we also overlay economic data at a post code level to estimate the 30-day risk of default (PD30). But now we have added in 90-day default estimates (PD90) and the potential value which might be written off, measured in basis points against the mortgage portfolio. We also calibrated these measures against lender portfolios.

So today we walk though some of the findings, and once again demonstrate that granular analysis can provide a rich understanding of the real risks in the portfolio. Risks though are not where you may expect them!

First we look risks by by state. This chart plots the PD30 and PD90 and the average loss in basis points. WA leads the way with the highest measurement, then followed by VIC, SA and QLD. The ACT is the least risky area.

So, looking at WA as an example, we estimate the 30-day probability of  default in the next 12 months will be 2.5%, 90-day default will be 0.75% and the risk of loss is around 4 basis points. This is about twice the current national portfolio loss, which is sitting circa 2 basis points.

Turning to our master household segmentation, we find that our Multicultural Establishment segment has the highest basis point risk of loss, at around 3 basis points, followed by Young Affluent, Exclusive Professionals and Young Growing Families. This immediately shows that risk and affluence are not totally connected. In fact our lower income groups, are some of the least risky. The PD30 and PD90 follows this trend too.

The Loan to Value bands show some correlation to risk, although the slope of the curve is not that aggressive, indicating that LVR as a risk proxy is not that strong. This is because in a rising market, LVRs will rise automatically, irrespective of serviceability.

A more sensitive measure of risk is Loan To Income (which APRA mentioned yesterday for the first time!). Here we see a significant rise in risk as LTI rises. Above 6 times income the risk starts to rise, moving from around 3 basis points, to 6 basis points at an LTI of 10, and 12 basis points at an LTI of 15+. So rightly LTI should be regarded as the leading risk indicator, yet many lenders are yet to incorporate this in their models. It is better because in the current flat income environment, income ratios are key.

Age is a risk indicator too, with households below 40 showing a higher risk of loss (3 basis points) compared with those over 50 (2.25). Even those into retirement will still represent some level of risk.

Finally, and here it gets really interesting, we can drill down into post codes. We plotted the top 20 most risky post codes across the country from a basis points loss perspective. What we found is that in the top 20 there is a high representation of more affluent post codes, especially in WA, with Cottlesloe, Nedlands and City Beach all registering. We also find places like Double Bay and Dover Heights in Sydney, Hinchenbrook  in QLD and Caulfield in VIC appearing. These are, on a more traditional risk view, not areas which would be considered higher risk, but when we take the size of the loans and cash flows into account, they currently carry a higher risk profile from an absolute loss perspective.

So, we believe the time has come for more sophisticated, data driven analysis of mortgage risks. And risks are not where you might think they are!

Digital Finance Analytics – Quenching The Thirst For Accurate Household Mortgage Data

Digital Finance Analytics Core Market Model is now being used by a growing number of financial services companies and agencies who want to understand the true dynamics of the current mortgage market and the broader footprint of household finances across Australia.

The DFA Approach

By combining our household survey data, with private data from industry participants as well as public data from government agencies we have created a unique statistically optimised 52,000 household x 140 field resource which portrays the current status of households and their financial footprint. Because new data is added to each week, it is the most current information available. We also estimate the extent of future mortgage defaults, thanks to the data on household mortgage stress.

Posts on the DFA blog uses data from this resource.  Momentum in our business has picked up significantly as concerns about the state of household finances grow and the thirst for knoweldge grows. We plug some of the critical gaps in the currently available public data which is in our view both limited and myopic.

A Soft Sell

The complete data-set is available purchase, either as a one-off transaction, or by way of an annual subscription which includes the full current data plus eleven subsequent monthly updates.

Other clients prefer to request custom queries which we execute on a time and materials basis.

In this video you can see an example of the core model at work. We show how data can be manipulated to get a granular (post code and segment) understanding of the state of play.  This is important when the situation is so variable across the states, and across different household groups.

We Hold Granular Data

  • Household Demographics (including age, education, structure, occupation and income, location, etc.)
  • Household Property Footprint (including residential status, type of property, current value of property, whether holding investment property, purchase intentions, etc.)
  • Household Finances (including outstanding mortgages and other loans, credit cards, transaction turnover, deposits, superannuation and SMSF, and other household spending)
  • Household Risk Assessment (including loan-to-value, debt servicing ratio, loan-to-income ratio, level of mortgage stress, probability of default, etc.)
  • Household Channel Preferences (including preferred channel, time on line, use of financial adviser, use of mortgage broker, etc.)
  • Segmentation (derived from our algorithms; for household, property, digital and others)

Request More Information

You can get more information about our services by completing the form below, where you can also request access to our Lexicon which describes in detail the data available.

You should get confirmation your message was sent immediately and you will receive an email with the information attached after a short delay.

Note this will NOT automatically send you our ongoing research updates, for that register here.

 

 

What The Mortgage Stress Data Tells Us

Following the initial release yesterday, and the coverage in the AFR, today we drill down further into the latest mortgage stress results.

By way of background, we have been tracking stress for years, and in 2014 we set out the approach we use. Other than increasing the sample, and getting more granular on household finance, the method remains the same, and consistent. We can plot the movement of stress over time.

Remember that the recent RBA Financial Stability review revealed that 30% of households were under pressure with no mortgage buffer, and a recent Finder.com.au piece suggested more than 50% were unable to cope with a $100 a month rise. So we are not alone in suggesting households are under greater financial pressure.

For this analysis we plot the number of households in mild stress (making mortgage repayments on time but tightening their belts so to do); severe stress (insufficient cash flow to pay the mortgage), and also an estimation of the number of households who may hit a 30-day default within the next 12 months. This is calculated by adding in a range of economic overlays into the stress data. This is all done in our core market model, which contains data from our rolling surveys, private data from lenders and other sources, and public data from the RBA, APRA and ABS.  This model is unique in the Australian context because it runs at a post code and household segment level, allowing us to drill into the detail. This is important because averaging masks significant variations.

The analysis shows that there are more severely stressed households in NSW than other states, and that around 13,000 households risk default in the next year, a similar number to VIC. WA is third on this list, with the number of defaults lower elsewhere.

Another lens is by the locations of households, in the residential zones around our major cities. The highest risk of default resides in the our suburbs, where a higher proportion of households are in severe stress. Households in inner regional Australia are next, followed by the inner suburbs, where again more households are in severe stress.

Our core household segmentation shows that the highest count of defaults are likely among the suburban mainstream, then the disadvantaged fringe, followed by mature stable families and young growing families. It is also worth noting that the young affluent and exclusive professional, the two most affluent segments contain a number of severe stressed households. This have larger mortgages and lifestyles, but not necessarily more available cash.

Finally, for today, here is the mapping across the regions. No surprise that the largest number of stressed households are in the main urban centres of  Melbourne and Sydney.

Next time we will look at post codes across the country.

 

More On Negative Gearing Distribution – The Wealthy Benefit The Most

Last week we discussed data from our core market model on negative gearing, and using our segmentation demonstrated that some, and more wealthy segments, benefit the most.  There is room to trim the excesses, without necessarily removing gearing overall.

Today we look at another perspective, which supports this argument. We estimate that 61.7% of households with investment property are negatively geared – this has been rising significantly, as investment property penetration as risen.  Around 2.4 million households hold investment property, but not all is mortgaged or geared.

The first chart shows the value of investment property mortgages mapped to the value bands of investment property held. The orange area are households who negatively gear, the blue those who do not. This shows that the larger value portfolios have more gearing, and therefore get the greater tax benefits.  Note also the small, but important peak in portfolio values above $2m. We are seeing the rise in the “professional” investor class, or Portfolio Investors as we call them.

Another way to look at the value distribution is by the number of properties held in the investment portfolio. Again the orange area is property negatively geared, the blue, not geared.  We see a significant spike in gearing above 5 properties, as well an an expected strong distribution in one or two properties. Our modelling shows around 79% of households have one or two properties.

The overall costs of negative gearing and capital gains tax concessions are an estimated $7.7 billion annually, and three-quarters of the capital gains tax concessions are enjoyed by the top 10 per cent of income earners.

So, in our view, the Government should be looking to curtail the gearing available to multiple property holders, and limit the total amount which can be geared. Those two simple measures would take heat out of the market, reduce the tax burden and still allow “mum and dad” investors to benefit.

A categorical “NO” to negative gearing reform is a major mistake. Treasurer, please note! As it stands, as mortgage rates rise, and investment loans will bear the brunt of these rises, actually the poor tax payer pays for this, insulating geared investors from the extra costs. Treasury should be modelling the extra impost this will be on the budget.

 

Investor Property Footprints And Negative Gearing

The argument trotted out to defend negative gearing from reform is that the bulk of investors are “typical mum and dad” households.

Of course it depends on how you look at the data, but lets look at output from our core market model.

What we have here is the relative VALUE distribution of investment property held by our core household segments, based on marked to market values.  We see that whilst some households in most segments are represented, the relative value is massively skewed towards more wealthy segments. Exclusive Professionals, our most wealthy segment holds 27% of all investment property by value, Mature Stable families hold 18%, Suburban Mainstream 15% and Wealthy Seniors 9%.

Another way to look at the data is through the lens of our property segmentation. Here investor only segments (they have no owner occupied property) hold 33% of investment property. Within that Portfolio Investors who hold multiple properties hold 3% by value. Those holding property but with no plans to move – Holders – have 20% by value, whilst those trading down hold 19%.

When we look at households by employment type, we see that employed workers hold 62% by value, whilst 17% are help by those not working, 10% managers, 9% expert professionals, and 2% by executives.

But if we look at the use of negative gearing, we see that three segments, by value have the largest footprint. Exclusive Professionals have 42% of negatively geared property, Mature Stable Families 27%, and Wealth Seniors 14%. Other segments are much less likely to negatively gear.

Looking again by Property Segments, Investors and Portfolio Investors have 32% of all negative gearing by value, but other segments also use this technique.

From this we conclude that it is important to separate the holding of an investment property from the use of negative gearing against that property. In fact we think negative gearing is predominately used by more affluent households, and they get the biggest tax breaks as a result, which of course other tax-payers have to subsidise.

There is, in our view, overwhelming evidence that curtailing the excesses in negative gearing (for example, a $ limit) would assist in cooling the market and inject needed cash into the budget.

But as we pointed out the other day, if the political agenda wins out, this just will not happen.

Property And Household Financial Footprints

Data from the Digital Finance Analytics Core Market Model tells an interesting story when we look at households dependence on wealth from property.

To illustrate the point, here are three charts, looking at different household groups. The first is the owner occupied mortgage group.

The blue area represents the distribution of households by age bands. The yellow line shows the relative value of total net worth (assets less debt, including superannuation). The green dotted line shows the value of property, in today’s terms, and the red line the current mortgage. It is very clear that older Australians have greater net assets and smaller mortgages. It is also clear that much of that worth is from paper profits relating to property. They would take a bath if prices were to fall.

Households without a mortgage have greater worth in other savings vehicles, including shares, deposits and property. They are more insulated from property value falls, and of course would not be hit by rising mortgage rates directly.

Finally, those who rent have a lower average net worth. Younger renters have little in the way of assets, whereas older renters on average hold higher balances, partly thanks to superannuation.

The analysis reconfirms how critical property values are to overall net worth. As a nation, we are highly exposed to future price movements. Any correction, whilst it might make property accessible for first time buyers, will seriously erode the net worth of households, especially those in the older age bands. The on-flow to economic outcomes suggests the risks are real, as Phillip Lowe said last night.

Half Of Households Are In Rental Stress

According to the latest modelling from Digital Finance Analytics, around half of all households in rental accommodation are struggling to pay their rent on time.

Across all households, more than 30% are renting, and this has been rising as the costs of property escalate, mirroring the rise in mortgaged households.

Within the rental sector, around half are fine, but 37% are in mild rental stress (meaning they are making their rental payments by cutting back on other spending, putting more on credit cards and generally hunkering down). An additional 13% are in serve rental stress (meaning they are struggling to pay their rent on time and are likely to fall behind). We look at total cash flow, not a set proportion going on the rent (e.g. 30%).

Static incomes, underemployment and rising costs of living all add to the pressure, despite an overall fall in rental yields.

There is a strong correlation between rental stress and the proportion income going to make rental payments.  In some cases of severe stress, there is not enough income from all sources directly to cover the rent, and they are forced to borrow to fill the gap, or use savings.

We can also look across the rental sector by our household segments. Seniors are most likely to be in severe stress, but other groups are also being hit by rental stress. Many wealth seniors are tapping into savings to survive but stressed seniors do not necessarily have this option.  Fuel bills are a particular concern for many.

This analysis shows that we cannot just focus on housing affordability for owner occupied purchasers; housing policy must also cover the rental sector, where the supply of affordable rental property is a major issue. Once again joined-up strategic thinking is required to tackle this intractable problem.