Tricky world of house price statistics

By Bruce Morris

Statistics are a huge part of the real estate game. But in the wrong hands, as Bruce Morris writes, they can camouflage the truth

Median data can be a trap. Photo / Thinkstock
Median data can be a trap. Photo / Thinkstock

If Mark Twain was around today, he'd be tempted to rejig his century-old endorsement on three kinds of lies first noted by British Prime Minister Benjamin Disraeli. To "lies, damned lies and statistics" he might have added house price data.

Statistics on everything from cricket to the weather are rich and meaningful if they are understood and handled with care but they can often distort reality if good judgement is missing.

And that is certainly so in the business of real estate. Don't blame the wider industry, however. The real problem is the interpretation (or lack of explanation) of the data once it gets into other hands.

A range of measures is fed to a hungry public from so many sources in the property game, at times leading to a confused picture.

One day the Real Estate Institute may reveal national figures, then it's one of the big industry players on its own sales, followed by state-owned valuation giant QV on all registered sales.

Depending on the cycle and the measure, the results can be quite different. The problem is not with the statistics - it's what's read into them.

The institute and QV (or its website owner, PropertyIQ) give regular results from their index systems to reveal how much values have risen or fallen in the past month, quarter or year.

At other times, along with the big agencies, they may turn to median and average sales or, in PropertyIQ's case, the in-house valuation tool E-Valuer to give an idea of how the market is moving.

Each system has its value and all are closely watched by the Government, Reserve Bank and economists. But they know what they are looking for and will not jump to silly conclusions; their interest is generally in overall trends across the country, cities and regions and not what may or may not be happening in a small suburb (though that will be of interest to local authorities when rating capital values are set).

Median data is a real trap. It can have strong value when spread across a region covering hundreds or thousands of sales, but it is often meaningless when applied to small areas. Even large areas can produce unusual results if the mix of properties selling changes considerably.

A couple of daft examples: the latest REINZ statistics show the median sale price in "Manukau Rural" (covering Clevedon, Kawakawa Bay, Orere and Orere Point) was $300,000 in December - down a staggering 43.4 per cent from the previous month. So does this mean values in rural Manukau are tumbling? Of course not. A slightly deeper look at the figures will tell anyone interested that there was just one sale in the area in December. A month earlier, there was also a single sale, and that one was a more expensive place.

In the Waitakeres (covering Cornwallis, Karekare, Piha and Waitakere) the median price jumped 36.6 per cent in December. But that leap was on the back of just five sales and no one should be imagining there's something crazy going on out beyond the hills.

The point is that median sale statistics (and average sales, for that matter) often give ludicrously skewed results and offer no guide to general value unless the volume is substantial. If more higher-priced properties sell one month, the median will leap; if the focus in another month is more on cheaper properties, the median will plummet.

In the end, it comes down to numbers. Even if the statistics cover a big suburb with 25 or 30 sales a month, that is scarcely enough to give precision on what is happening generally in that area, which is why many commentators go back three months (and perhaps 75 or 100 sales) to get a reasonable fix. And even then they tread warily.

The same caution and reserve is often missing in the media. How much of the air in the latest property bubble in Auckland is due to the endless stories on catapulting prices and $1 million do-ups, with little to balance the risks of burnt fingers when mortgage rates rise and other moderating factors start to bite? If media stuck with the broad picture when using median data, they would better serve their readers and listeners.

Who, for example, could have blamed naïve readers for imagining that country towns are the places to go and buy a property after reading the front page of the Sunday Star- Times in the week before Christmas?

The spread, taking the whole of the front page, was headlined "Real property hotspots - NZ's high-growth towns and suburbs". It was built from a table of "20 real estate hotspots" identified by median price movement in the year to the end of November.

The prominence given surely left many readers in no doubt that values had soared in that year - yet anyone with even a vague interest in New Zealand residential real estate would know that Auckland is surging and rural towns have been doing it tough, and have been for the past five or six years.

Astonishingly, Mt Roskill was the only Auckland suburb on that reputed hotlist - when any decent analysis would have the city holding every spot.

It was a staggering piece of misinformation, hardly helped by the reporter's declaration: "The figures do not appear to be an anomaly."

Top of the boomtown list was Foxton where the median price had leapt 59 per cent over the 12 months and every indication was given to readers by the prominence of the spread that general values had risen by that sort of figure.

But it was rubbish, and illustrates how false impressions can be created by relying on median statistics without further research and sensible qualification.

Rather than stand out as a real estate hotspot, Foxton was in fact one of the country's worst performers in 2012.

Using its E-Valuer - a refined and respected valuation tool giving constantly updated values focusing on similar local sales and other measures - PropertyIQ tracks the value of all properties in an area, not just those that happen to have sold.

By this measure, Foxton was near rock bottom in the residential property performance league. Of the 826 New Zealand suburbs and towns with enough sales to allow an assessment of average value, Foxton sat at number 820 through the survey period.

At the end of November, PropertyIQ assessed average residential property value in the town at $128,333, down by nearly 9 per cent over the previous year and 22 per cent off the highs of six years ago. So much for hot-as-hell Foxton.

Every town and suburb on the list did much worse than the newspaper "research" suggested. Rather than rise in value to "hotspot" status in that year on the basis of median selling price, a number joined Foxton in showing actual decline, if PropertyIQ's valuation tool is accepted as a fair and accurate way of measuring average value.

Again, the problem here wasn't with the data itself. Foxton's median did rise 59 per cent over the 12 months, but that reflected a very low 2011 base in sales and price and more sales of high-priced properties. It did not mean overall values had risen by a similar amount - yet that was the message sent to gullible readers.

Property Report concentrates on average value assessed by E-Valuer in its centre data pages covering 420 North Island towns and suburbs. One data column also provides detail on quarterly selling prices against actual rating capital values, giving property owners a rough guide on what's happening on their patch. But call real estate agents and valuers for a true market price on individual properties.

- NZ Herald

© Copyright 2014, APN New Zealand Limited

Assembled by: (static) on red akl_n2 at 11 Jul 2014 20:34:52 Processing Time: 1610ms