Infometric's Gareth Kiernan takes a look at where mortgage interest rates are headed in light of the recent changed signals from the RBNZ

By Gareth Kiernan*

Since late 2022, we’ve been predicting that mortgage rates would peak at over 7% this year based on our outlook for wholesale interest rates. But since the start of 2023, mortgage rates have generally tracked sideways or downward. Why are financial markets seemingly resistant to the Reserve Bank’s efforts to quell inflation by weakening demand through tighter monetary conditions?

How do we forecast mortgage rates?

The foundations of all our interest rate forecasts are two key rates: the official cash rate (OCR) and 10-year government bond rate. The OCR reflects expectations of policy decisions the Reserve Bank will take to meet its mandate, which is primarily to keep inflation between 1% and 3%pa, and the OCR is highly correlated with wholesale 90-day bank bill rates.

The connection between the OCR and 10-year bond rates is more nuanced. In theory, bond rates reflect expectations of short-term interest rates over the life of the bond. But there are times when OCR increases can lead to lower long-term bond rates, if financial markets believe the Reserve Bank is overdoing its tightening now and will need to cut the OCR further in future, for example. And with trends in international bond rates also having a bearing on New Zealand’s bond rates, our bond rate forecasts have a high degree of independence from our OCR forecasts.

Our fixed mortgage rate forecasts are a weighted average of the 90-day and 10-year rates outlined above with an additional “premium” built in for the retail margin. Our estimate of the premium is reasonably dynamic and based on recent market trends, which we allow to revert towards a longer-term historical average over time. We also recalibrate the model’s weightings of 90-day and 10-year rates for each fixed mortgage rate from time to time.

We have looked back at our modelling over the last 16 years and pulled out five representative sets of 90-day and 10-year weightings for fixed mortgage rates. In Chart 1, we have plotted the two-year mortgage rate against the highest and lowest modelled two-year rates. The trends for the two-year rate are mirrored across other fixed mortgage terms.

The need to align the maturity profiles of funding and lending

Prior to 2009, two-year mortgage rates usually tracked around the bottom of the range in Chart 1. The catalyst for a shift around 2009, which broadly coincided with the Global Financial Crisis, was a change in macroprudential requirements from the Reserve Bank that required lenders to match up the maturities of their lending books with their wholesale funding. Previously, for example, banks had been able to offer a two-year mortgage rate based on their “bets” of where wholesale rates would go over the term of the mortgage.

So, if Bank A thought the Reserve Bank was going to cut the OCR by more than financial markets were expecting, it could take a gamble to try and increase its market share and offer a lower two-year mortgage rate than Bank B, whose pricing would be more in line with two-year wholesale rates. If Bank A’s forecasts were correct, it could instead roll over short-term wholesale funding throughout the two-year mortgage term and maintain its profit margins as the OCR fell by more than markets had been expecting. If Bank A’s forecasts were wrong, its margins would be squeezed.

Even before the Global Financial Crisis threw much of the world’s banking system into turmoil, the Reserve Bank had adjudged that the risk of these bets going wrong and causing significant bank losses was too great – hence the change in requirements outlined above. The subsequent alignment of banks’ funding and lending maturity profiles appears to have added about a percentage point to the “premium” for the retail margin. As a result, the two-year fixed rate has tended towards the top of our forecast range between 2009 and 2020.

The temporary deviation from this trend in 2020/21 is likely to have been caused by the provision of cheap funding through the Reserve Bank’s Funding for Lending Programme. But the deviation since the start of this year has no such obvious cause. If anything, we would have expected the gradual introduction of increased capital holding requirements by the Reserve Bank (a change that was postponed due to COVID-19) might be starting to drive another step increase in the “premium” built into banks’ lending rates.

Swaps and discounting not the whole story either

Banks typically use swaps to convert short-term funding to longer-term funding to align with their lending maturity profile. These wholesale swap rates move closely, but not exactly in tandem, with government bond rates. Since about November last year, swap rates have trended lower than bond rates, implying that banks have experienced less upward pressure (or, in the case of five-year rates, more downward pressure) than government bond rates have otherwise suggested.

Although swap rate movements might offer some explanation for mortgage rate trends this year, it’s not an entirely compelling one. The current gaps between swap rates and bond rates are not particularly different to their long-term averages, so the fact the gaps have shrunk over the last six months doesn’t explain why fixed mortgage rates have broken away from the top end of our modelling range.

There has also been evidence of some mortgage rate discounting taking place by banks this year as they chase market share. Exactly why banks would be falling over themselves to lend money on an asset class that is declining in value is unclear, but when turnover in the housing market has almost halved over the last two years, lending targets imposed by sales managers will have become increasingly difficult to meet. February saw BNZ offering a “secret” 4.99% one-year rate, with less aggressive specials of 6.19% for 18 months and 5.99% for three years also appearing over the last few months.

But even these selected examples fail to explain the divergence of fixed mortgage rates from our modelled trend shown in Chart 1, which is repeated across all fixed mortgage terms. Discounting of a specific fixed mortgage rate to attract more borrowers would only show up for that selected lending term, rather than across all fixed rates.

Don’t fix long at the peak

Back to the original question: much of the above discussion is moot considering last week’s Monetary Policy Statement, where the Reserve Bank signalled that the OCR has probably reached its peak. The lift in wholesale interest rates following the government’s expansionary Budget was reversed out by the Bank’s Statement, and financial markets are now firmly thinking about when the Reserve Bank could start to cut the OCR. Although the Bank’s own forecasts suggest a cut might not occur until the second half of next year, longer-term wholesale rates are likely to gradually factor more cuts into their pricing – particularly if headline inflation continues to track downwards as forecast.

Perhaps the most important lesson to remember is the one learnt by borrowers in 1998 and 2008: jumping on long-term rates because they’re the lowest currently on offer can come with significant costs later. In April 1998, five-year fixed rates were at 9.4%, compared with one-year rates of 10.0% – but by April 1999, one-year rates had plunged to 6.1%, so being locked into that 9.4% for another four years was eye-watering. Similarly in April 2008, five-year rates were 9.5% and one-year rates were 10.7%, but the latter had dropped to 5.8% a year later. With five-year rates currently sitting at around 6.4%, the potential for regret now is nowhere near as large. But if this is the peak in mortgage rates, it might the last point in time you want to be fixing your mortgage for too long.

*Gareth Kiernan is chief forecaster at Infometrics. The original article is here.