Median house and unit prices for a suburb will tell you what the ‘middle’ sales price has been over a period of time, such as a year. A suburb may have 100 sales in a year; the middle price is halfway between the 50th and 51st sale ranked in order of price. Changes in the median price data over time is used by many people to determine changes in asset values, and is often cited to justify why you should buy in a particular suburb (it has grown x % in the last 12 months).
Leaving aside the obvious point that past performance is not an indicator of future performance, there are some fundamental issues with median prices as a statistic which you should be aware of. Like any statistic, its usefulness is determined by:
- The amount of data you have
- The time period you have the data over
- The level of variability in the input data
In the ideal world, we could track median sales prices with lots of data over an unlimited timeframe with no variability in the inputs. It would look something like this. We track the price of a suburb with a thousand sales per year, over a 50 year period of time. All of the properties in the suburb are identical in attributes, and no owner improves the value of their property in any way. This would give us an extremely accurate picture of median sales price movement. The amount of sales data would exclude sales data which is misrepresentative, the length of time the data was acquired over would give us a true picture of performance over a genuine investment timeframe, and the consistent input data would allow us to make accurate deductions about what the data meant to all of the properties in the suburb.
Clearly though, this isn’t the case. Some suburbs have very few sales, or sales fluctuate dramatically between years. Much of the median sales price data is given to you for a paltry 1,2 or 5 year timeframe – well below a true investment timeframe. Finally, the data we have varies considerably; while some properties may have changed very little over a period of time, others will have been completely renovated and extended. In suburbs with high levels of development, new properties will often push up the median price without affecting (or in many cases reducing) the value of existing property. There’s a great article from Michal Matusik which highlights this - read it here.
An example of North Lakes in northern Brisbane is a good one. We can see here that the suburb didn’t have a meaningful number of sales until 2000-2001, which gives us 18 years of useful data. During this period there’s been a big increase in sales as the suburb expanded, with a slight tail off lately as development has slowed. And what about price growth? There’s some very strong growth from 2000-2008, which is too high to be purely written off as the increase in the cost of new housing. However, from 2008-2017 (the last full year of data), growth has been only $43,500, or just over 1% per year. In reality, property built 10 years ago is selling today for roughly what it cost to buy, and in many cases below this. So the small 1% per year gain is a reflection of newer housing pushing up the median, rather than a genuine gain for the unfortunate investors of 10 years ago.