ANZ Bank has looked at how nearly all economists and the Reserve Bank got house price predictions so wrong since Covid snuck into New Zealand last year. 

Rewind the clock a little more than a year, and, for example, the ANZ’s forecast for double digit house price drops stand in stark contrast to the 30% plus year-on-year house price inflation that occurred.

Even the Reserve Bank has admitted its house price forecasts have been consistently wrong in the past decade.

The bank, told a parliamentary select committee earlier this month its forecasts had been off by an average of 5.2% since 2010, with house prices rising consistently higher than forecast.

It blamed higher migration and lower mortgage rates for its wrong forecasts.

So why were ANZ, the Reserve Bank and just about every other forecaster so wrong about the house price outlook?

Underlying factors

ANZ chief economist Sharon Zollner says the answer lies in how the underlying factors influencing housing evolved relative to expectations at the time, and how the macroeconomic and Covid health policy response turned out to be a much stronger stimulatory cocktail than anyone imagined.

During the first Covid lockdown and afterwards business and consumer confidence fell through the floor.

Commodity prices – a signal for global demand – plummeted; the country’s biggest export earner, international tourism, was closed for business for the foreseeable future; and a hefty chunk of the economy had just been switched off under one of the world’s most stringent lockdowns.

Zollner says the income shock associated with this looked like it was going to be the largest seen in a generation.

The view at the time was the business cycle needed all the help it could get to offset the income shock and prevent a blowout in unemployment.

“All the while, the possibility of frequent and/or prolonged alert level restrictions – along with the assumption they were more economically damaging than now believed – hung over the economy’s head in such a threatening fashion the prospect of a recovery in business confidence seemed slim.

“As the country now knows, the economy’s evolution has been much less dire than all this.”

Perhaps the main ingredient of the recovery that was underestimated was just how successful New Zealand would be at containing outbreaks and avoiding lockdown, says Zollner.

That allowed for higher-than-otherwise economic freedom and mobility, boosted confidence, and made the transmission of fiscal and monetary policy much more powerful than assumed.

“Once momentum changed direction, there was so much volatility in the data and lingering virus uncertainty that it took a while for economists and government institutions to centralise it in their outlooks.

“We now find ourselves in that same situation again, where much depends on just how successful the country is at eliminating Covid, and the data is going to be all over the show.”

Models can’t be the answer

Zollner says when it comes to forecasting house prices there is certainly no single model of “truth”.

“All models are nothing more than an extremely simplified representation of a market, and what is put in largely determines what you get out.”

She says it’s important to remember that while some models might work well most of the time, they are probably going to be pretty rubbish when the “big stuff” happens, which is precisely the time when everyone wants accurate forecasts.

“The correlations between variables you’re taking advantage of might not be all that stable in such conditions.”

In other words, models of the market are not the market and economists need to maintain a healthy scepticism of the forecasts they are producing, particularly during a significant shock such as Covid-19.

It is easy, says Zollner to develop house price models based on economic forecasts of key variables such as population growth, interest rates, income growth, and growth in the number of houses, for example.

But forecasts of these variables are likely to be wrong to some extent, and more likely to be more wrong when done in the middle of a crisis.

There are many other forces influencing housing that are not easily quantified, such as “animal spirits”, macro-prudential policy settings – eg LVR restrictions – tax policy changes – the removal of interest deductibility for investors – and compositional changes, such as the dominance of returning New Zealand citizens in recent net migration, she says.

“While economists do attempt to isolate and estimate the impacts of these harder-to-quantify influences, there is no way to fully capture them in the data or consider them together in a statistically robust model.

“That means we need to lean on supplementary analysis to inform our qualitative judgements that overlay our model-based forecasts – and each time we do this introduces new assumptions and uncertainties.”

Zollner says knowing what’s in the forecasting sausage hopefully helps demonstrate just how uncertain anyone’s house price forecast is, particularly at times like these.

“No-one has perfect foresight, but, dangerously, 20/20 hindsight can create that illusion, and no-one has a perfect model of the real world.

“As long as decisions that affect tomorrow need to be made today, there will always be demand for forecasts.”

She says an economist’s job is not to “sell” the one true outlook. It’s to present a central view, highlight the assumptions that feed into it, and express the uncertainties around it to help people make better informed decisions.