Historic drop in house prices ahead

The new year is playing out as expected. Central banks are, for the time being, clinging to their hawkish rhetoric that further rate increases will be required to crush the inflation crisis.

They refuse to seriously countenance the risks of over-tightening policy, explicitly favouring erring on the side of lifting rates too far over the alternative of allowing consumer price pressures to become entrenched.

This is terrible news for asset prices, which is why US equities are already trading almost 1 per cent below their December 30 marks. And it only raises the already elevated probability of both a US and a global recession.

An emboldened US Federal Reserve, cock-a-hoop from the recent roll-over in the headline and core inflation data, has been at pains to dismiss rampant speculation of interest rate relief in 2023 notwithstanding the bond market pricing in 33 basis points of cuts before the year is out.

The message is that interest rates are to remain highly restrictive until there is overwhelming evidence to convince policymakers that core inflation will converge to their circa 2 per cent targets.

And yet, the global inflation data continues to surprise on the downside, with the latest European prints undershooting market forecasts. Activity data has also been mixed, and will only continue to sour.

This has helped long-term government bond yields decline sharply this month after a sustained increase at the end of last year. Tellingly, the current inversion in the US yield curve, which reflects the fact that the interest rate on the two-year US government bond is higher than the 10-year bond, has increased markedly since last year.

More precisely, the 10-year US government bond yield is trading an extra 20 basis points below the two-year yield. In numerical terms, the US two-year government bond yield is 4.46 per cent, 74 basis points above the 10-year yield of 3.72 basis points. This is close to peak bond market negativity. You have to go all the way back to 1981 to find a US yield curve that was more inverted. And this inversion is typically a very good signal of an impending recession.

Aussie housing

In Australia, housing valuations continue to fall at a rapid clip. Based on the latest CoreLogic data to January 6, national capital city prices have slumped more than 9 per cent since their May 2022 peak, which is not far off the historical record drawdown of 10.7 per cent between 2017 and 2019 (using the publicly available five capital city index).

On our estimates, Aussie housing should register its worst cumulative loss in 42 years within two to three months. In Sydney, home values have plunged more than 13 per cent while Brisbane is not far behind, chalking up a 9.7 per cent decline. Those modest Melburnians have only slightly outperformed the other capitals with an 8.5 per cent correction thus far.

I was recently asked how the housing market’s evolution is tracking against our October 2021 forecast for a total peak-to-trough correction of 15 to 25 per cent assuming the Reserve Bank of Australia raises interest rates by at least 100 basis points, which we reiterated in November 2021.

The bottom line is that the market is behaving as expected: we are now 60 per cent along the road to hitting the lower bound of our projected loss. To the best of our knowledge, no mainstream analysts were forecasting any house price losses – let alone a record drawdown – in October 2021.

The singular driver of this heterodox perspective was our belief that the RBA would seek to normalise its target cash rate in 2022, which would radically reduce buyers’ purchasing power. On June 3, 2022, one month after the RBA began lifting its cash rate, this column declared that “the great Aussie house price correction has begun”.

Improving the RBA's housing model

There has been confusion over our reference to an assumption of at least 100 basis points of RBA interest rate increases, which some took to be a projection. In October 2021, we did predict that the RBA would begin normalising rates in the second half of 2022. As it happened, it initiated the process in May 2022.

The 100 basis point interest rate increase assumption derived from our application and refinement of the RBA’s complex and innovative model of the housing market, developed by Trent Saunders and Peter Tulip when they worked at the RBA. As we have explained before, this model controls pretty much all the demand- and supply-side factors that you can imagine influence house price movements, including mortgage rates, the “user cost of housing”, rents (where rents are driven by population growth), building construction, household income and unemployment.

In 2021, Coolabah hired Tulip as a consultant to work with us on refining the model, including ironing out some wrinkles that our chief macro strategist, Kieran Davies, had identified, as well as discussing macroeconomic inputs into the model.

The Saunders/Tulip model allows one to examine the impact of permanent and temporary changes in interest rates on house prices over time. In 2021, the RBA’s cash rate was sitting at its record low of 0.1 per cent. Our initial approach was to quantify the impact of a permanent increase in the cash rate of 100 basis points. This generated a correction in house prices of slightly more than 30 per cent.

Shortly after the RBA started raising rates last May, we published an additional note outlining our modelling of a more dynamic RBA cycle that involved it lifting its cash rate from 0.1 per cent to a peak of 4.25 per cent in 2023 then cutting rates in 2024 and 2025.

This furnished a slightly more pessimistic 30 to 40 per cent total peak-to-trough correction in national house prices over the forecast period.

While we have been historically happy to publish forecasts of house price changes, we are not in the business of releasing interest rate projections (despite being often asked to do so).

Our central case unveiled in October 2021 was notably more optimistic than the outlook presented by the Saunders-Tulip model because we adopt a “quantamental” approach. Pretty much every bank in the country has since embraced this view.

Indeed, the RBA revealed in recent freedom of information disclosures that it has spent a great deal of time trying to understand our application of, and refinements to, the Saunders-Tulip model. When the RBA started lifting interest rates in May 2022, its own internal use of the Saunders-Tulip model implied that Aussie house prices would not fall materially in the face of at least 250 basis points of cash rate increases.

This was, unsurprisingly, the worst calendar year result in 42 years. For better or worse, there is much more to come.

This was very significantly the RBA’s central case when it kicked off its monetary policy tightening cycle. Inside the central bank, a number of different economists tried to figure out how we were getting 30 per cent house price declines using the Saunders-Tulip model when the RBA’s own application generated no material reduction. Eventually, they were able to reconcile the differences.

Around the same time, the RBA was compelled to dump its May 2022 house price forecast for no significant fall in prices, and replace it with a new outlook that anticipated an 11 per cent correction, which it noted would be the largest since records began. This forecast is likely to be redundant within a few months, and the RBA has doubtless since replaced it with a new projection that is more in line with our larger, 15-25 per cent draw-down.

In October 2021, we thought the RBA would knock out two to three interest rate increases in 2022, worth up to 75 basis points. While we did not publish a specific forecast for house prices in 2022 (since we were only interested in the total correction), a substantial price fall was, by definition, our central case for 2022 along the road to our expectation for a cumulative 15 to 25 per cent decline.

And that is what we got with CoreLogic’s five capital city index declining 8.8 per cent from its May 2022 peak to December 30, or 7.1 per cent in calendar year terms. This was, unsurprisingly, the worst calendar year result in 42 years. For better or worse, there is much more to come.

Originally published in the AFR.