The CoreLogic research team has been forming a real estate sales volume prediction model to provide a guide to future market activity.

The data we entered ranged from GDP to unemployment, interest rates, wages and migration. And the results are pretty good.

The good thing is that where the predicted model diverges from actual sales, it's generally explainable. The single biggest occasion is in 2012, following the Christchurch earthquake, and is also the early part of the most recent growth phase in Auckland.

My feeling on why the model missed the increased turnover in Auckland is that it's because the exceptional lift in sales volumes at this time was heavily credit-driven (not interest rate-driven) which we didn't, and couldn't capture.


The model also misses the significance of the most recent slowdown, since 2016. Again this has been impacted by credit policy (outside of interest rates) and also the stricter LVR limits implemented at the end of 2016.

Understanding the blind spots of the model is crucial to being able to apply some form of manual adjustment to the final projections, based on all the other market influencers we're aware of. First though, the model results.

Based on forecasts by RBNZ and MBIE regarding the outlook for GDP growth and interest rates being stable, wage growth is anticipated to improve slowly and migration to drop slowly, we get to a prediction of sales volumes staying at about the same rate they're at.

Super boring, when we know extreme opinions sell, but that's what the model tells us.

The end result? Roughly 83,500 sales this year and 82,500 in 2019. This is down from 85,000 last year and 106,000 sales in 2016.

But now we apply our "adjustments", or alternative scenarios. The overwhelming potential pressure is downwards, with a number of measures reducing demand, either intentionally or otherwise. We have:

• The foreign buyer "ban"
• Healthy Homes Guarantee Act
• Ring-fencing of tax losses
• LVR limits
• The extension of the bright-line test
• And maybe debt-to-income ratios.

There are a few things which may increase demand, including the possibility of shared equity schemes to help first-home buyers and the potential relaxation of the LVR limits, but it's unlikely these will do enough to counter the downward impact of those listed earlier.

So, with so much downward pressure, the question is whether there will be a more significant drop in sales volumes (and consequently values).

But the good news for those in the industry is that a lack of supply, high migration (despite it reducing), historically low interest rates and an ingrained mentality of property investment all provide a solid floor supporting the market (both volume and value wise).

We can hypothesise different scenarios. For example, we estimate that a gradual lift of 1 percentage point in mortgage interest rates by the end of next year could see a further reduction in volumes of roughly 7000 to 75,000 in 2019.

This would be lower than anything we've seen in the past 20-odd years, and then probably be accompanied by further intervention to push things in the other direction (or at least hold things up).

The reality is that almost every day there's a new announcement or piece of information that could influence the property market, and a lot of people have a stake in it.

Having solid data to help us sift through the conjecture and potential bias is crucial, and our first sales projection model provides a great head-start.