We have written a few blogs on our view of residential investor mortgages and want to reinforce a few further points:
- Many banks view investor loans as lower risk than owner-occupier. This is wrong.
- Serviceability criteria will slow lending dramatically. ASIC is as much in-play as APRA.
- Serviceability is a yes/no result, meaning many investors will soon be blocked out from the market.
Investors loans have higher risk:
Most secured loan products for IRB banks follow the Basel PD, LGD formula to determine an unexpected loss value that is then a direct input into regulatory capital. The PD in this case is seen to be a through-the-cycle probability of default (TTC PD), which is then scaled up to a 1-in-1000 year stressed probability of default.
Australia has had benign lending environments in the last 2 decades, in effect no true credit cycle. So the first point to make is the TTC PD estimation is potentially a little low. In this data set, investor loans are showing a lower TTC PD than owner-occupiers*. One rationale banks believe is that rental properties have a replaceable occupant, so if one tenant loses his job, you can find another with a job.
However, the PD/LGD model is probably the wrong assessment for this asset class. Investment properties are revenue producing real estate, so you must wonder why residential properties get less stress than commercial property that gets put into Specialised Lending.
The PD scale-up is missing a systemic risk component that is not a spike in landlord defaults. It is an increase in rental property vacancy ratios.
Look at vacancy rates:
If an investment property is reliant on a tenant to cover the mortgage, usually interest-only, and the landlord doesn’t have a high capacity to cover the full mortgage payments themselves, then empty properties may lead to defaults.
So rather than looking at TTC PD, we should look at potential increases in rental property vacancy rates. Does a national movement from 2.5% to 12.5% seem appropriate for a 1-in-1000 risk event? We have seen unemployment rates move from 6% to 16%, so there is scope for at least a 10% increase in vacancy rates.
There is potential further stress in that a country-specific recession will have lower inbound migrants seeking work and expat communities might decline. Vacancy rates could increase further.
Added to this, rental prices will need to fall, due to excess supply of rental properties over demand. This is an unknown factor but will lead to further stress. As most investment property rental yields are below the mortgage interest rate, this is a dollar cost to the landlord’s income.
The owners of empty rental properties then face a dilemma. If they have a mortgage or higher loss to service, how will they pay for it? Relying on selling the property is not appropriate as 10% of other rental properties may be performing the same process at the same time.
Instinctively, 20% of investment properties becoming under severe stress feels about right. This is double the implied stress currently assumed on investor loans.
Serviceability is everything:
The above methodology is appropriate if the landlord is reliant on the tenant to cover the mortgage. If the landlord has a diversified source of other income, notably not just investment property, then you can assume they have the capacity to cover a single empty property.
The major issue with current serviceability calculations is they haircut the attainable rent by 20%. But as we described above, there could be an extra 10% of properties with no tenant and this is where the systemic risk is.
We expect ASIC and APRA to realise this risk at some point and tighten up the serviceability calculations so that investors reliant on filling a property will either not pass the serviceability assessment or they will be classified as Specialised Lending, which will have risk-weights 2 or 3 times higher and a mortgage rate several percentage points higher than owner-occupier home loans.
We look forward to these changes.
*Those that suggest this means there is lower risk in investor loans are incorrect. If expected losses do not appear, this is really a higher net income history rather than lower risk. Unexpected losses are the risk factor.