As rates reach their peak, should APRA reduce the serviceability buffer?

The conservatism buffer is a key regulatory setting that affects how much Australians can borrow, and thus shows what happens to house prices.

In late February, APRA left the buffer unchanged at 3 percentage points (ppt).

However, many, especially lenders, are asking for it to be reduced.

To understand what that means for borrower resilience, it’s worth looking at the data. Why it matters: A reduction of 0.5ppt increases borrowing capacity by approximately 5%.

Conservatism buffer ensures financial resilience from rising interest rates

The conservativeness buffer is the amount lenders add to current mortgage rates before assessing how much to lend to borrowers.

New owner occupiers, who are currently facing interest rates of around 5.5%, will have to be able to afford to repay their loans if interest rates rise to 8.5%.

Serviceability buffers are designed to insulate borrowers from interest rate risk. If interest rates rise, borrowers should be able to afford to repay their loans as long as the rise is less than the buffer.

Without a buffer, an increase in interest rates would force anyone who borrowed the offered cap to a position where they could no longer afford to pay it back.

However, most borrowers are more resilient than simply explained by the minimum buffer, as few borrowers owe the maximum amount (about 10% according to the CBA), according to the CBA.

A simple APRA change could ease some of the pain of rising interest rates, but is it worth it? Photo: Getty

---

Buffers and other serviceability settings have been significantly adjusted

Here is a rough recent history of buffers and related measures:

• December 2014: APRA standardized serviceability assessments requiring a minimum 2 ppt buffer and a 7% floor (the lowest serviceability rate regardless of current interest rates).[1]

• July 2019: Removed floor and increased buffer to 2.5ppt
• November 2021: Buffer increased to 3 ppt

For most of the last decade, the 7% floor rate was binding. That is, borrowing capacity did not change as interest rates changed.

But if interest rates were consistently below that level, “the gap between the 7% floor and the interest rate actually paid would be very large in some cases, perhaps unnecessarily.” According to APRA.

In other words, the amount of resilience built into mortgages was too great and imposed a heavy price in terms of limiting borrowing and keeping people out of the housing market.

Current buffer covers about 95% of historical interest rate rises

The buffer protects against interest rate risk, i.e. interest rates rising after the borrower has taken out the loan.

Since interest rates can rise indefinitely, there is no buffer that can eliminate all interest rate risk.

However, the first five or seven years of a loan are when borrowers are most vulnerable to rising interest rates. This is because after this point they have typically seen enough income and asset growth to significantly ease their mortgage repayment burden.

Borrowers rarely face mortgage rates that are 3 percentage points higher than their original interest rates. Such an increase is due to loans made from 2001 to his 2003 (which almost only affected his loans older than 5 years) and, most notably, loans made in 2021. only occurs.

Quantifying this interest rate risk, 95% of the interest rate increases during the first seven years of the loan life were less than 3.05ppt. This covers the last 30 years as the RBA has targeted inflation with interest rates.

This is almost exactly the current buffer level of 3ppt.

Alternatively, if we are only interested in the first five-year risk of a mortgage, 95% of the increase is less than 2.92ppt. not very similar.

As interest rates rise for the remainder of 2023, these distributions will shift slightly upwards. However, 3ppt seems about right to protect borrowers from about 95% of historical interest rate risk.

Lowering the buffer to 2.5ppt (the level before late 2021) would cover only about 80% of the borrower’s historical rate hikes in the first seven years, and only 90% in the first five years.

APRA should determine if this level of resilience is sufficient before making any changes to the buffer.

Adjusting the buffer based on expected interest rates is impractical

Intuitively, interest rates have risen sharply, so the buffer doesn’t need to be that big. Alternatively, the buffer could be reduced as rates are unlikely to rise by another 3ppt from now.

Similarly, when rates were low, they were more likely to “normalize” or increase. Broadly speaking, this was the motivation behind the 7% minimum maintenance ratio or floor prevalent in 2019 and earlier.

These ideas suggest that the buffer should be modified based on current interest rate levels. However, it is difficult to operate this in practice.

APRA requires accurate forecasts of interest rates. However, financial markets have been shown to have large forecast errors.

The reintroduction of floors also requires APRA to understand the ‘normal’ levels of mortgage rates. This is considered very difficult to identify and suggests that estimates have changed significantly in the past. In practice, lenders continue to use floor rates, and APRA needs strong evidence that floor rates are not set high enough.

Buffer settings could not be changed as quickly as interest rate forecasts, which are often sharply adjusted over short periods of time. And perhaps more importantly, if these projections are wrong, borrowers’ resilience could be compromised.

These are arguments against analytically well-defined buffers that remain constant throughout the economic cycle.

Buffers cannot explain inflation or income losses

APRA says: Conservatism is not only about interest rate risk, “The buffer provides a significant contingency not only for interest rate increases over the life of the loan, but also for unexpected changes in the borrower’s income and expenses.”.

In practice, reserve cash flows such as those provided by buffers can protect against these shocks, but buffers cannot be adjusted to account for these risks.

No loan is created to withstand a significant loss of income, such as unemployment.

Also, incorporating inflation into the calculation of the buffer makes it smaller, not larger. This is because income growth will almost always outpace spending inflation (although the current situation is a clear counter-example).

I don’t know yet if the buffer will be reduced

APRA needs to determine whether reducing the buffer to 2.5ppt and increasing borrowing capacity by 5% is worth reducing future borrower resilience from 95% of historical interest rate risk to 80-90%. I have.

However, adjusting the buffer based on expected interest rate movements is likely to be difficult in practice, requiring ongoing discussion and adjustment of appropriate serviceability settings. However, such settings are not designed to be changed on the fly.

It also puts a lot of pressure on getting interest rate forecasts right, so borrowers don’t run out of buffers when they need it most.

[1] In practice, these were thought to be minimal, so floor assessment rates were typically 7.25% or higher.