League tables for suburb performance based on medians can make for popular reading. However the results should often be challenged, especially where sample sizes are small, property sales not representative or where the price ‘distribution curve’ is skewed.
For an agent working in one of these ‘worst’ suburbs, it is always good to know the ‘true’ growth figure, especially when questions or concerns are raised by vendors or purchasers. Sometimes the truth may be that one of your local suburbs is performing poorly, however if it is the result of some statistical anomaly, you certainly do not want to be troubled by it when already facing tough market conditions.
Let’s look at two recently published ‘worst’ suburbs:
Bellara 4507 QLD Units
The published figure states a ‘minus 65%’ growth rate. The suburb has around 10 sales per quarter currently, giving us a median of around $116,000. Back in Q2, 2010 we saw just 4 sales, which returned a median of $360k.
Neither the current samples of 10 or the prior sample of 4 are large enough to rely upon. If we changed our sample time period from 3 months to 6 months, we see an entirely different result. Using 6 months to October 2011 gives us a median of $115k and a sample size of 15. The same period 12 months earlier returns a median of $110k and a sample of 15 as well. The growth rate using this approach returns a plus 5% result. This result is very different from the minus 65%.
Whilst the sample size of 15 is much more reliable than the quarterly sample size, the reliability of the data should probably be questioned due to the ‘distribution’. Using a ‘price segmentation graph’ we would prefer to see something that looks a little bit like a bell-shaped curve in the current sample and also the prior year sample. Without both periods offering us an approximately normal distribution of values, we can’t rely on the plus 5% result.
East Melbourne Houses
The league table states this suburb has fallen in value by 43% in the last 12 months. When we look at the total number of sales in the last 6 months (to October 2011) we count just 2 sales! It is hardly a representative sample size. Even if we expand the sample period to 12 months till Oct’ 2011 we still only have 8 sales.
Using the full 12 month sample period returns an entirely different result. The median for the last 12 months was $1.88m. Compared to the year prior (median of $1.67m) we can arrive at a positive growth rate of nearly +13%.
As per our Bellara sample, the price distribution hardly resembles a bell shaped curve and indicates that the result may be unreliable.
Tips:
Whilst by no means an exhaustive list, here are a few tips for agents when reviewing median prices.
- As a general rule, try and work with sample sizes of around 20
- If comparing time periods, ensure you maintain apples with apples (e.g. 6 month sample periods)
- Look at the price segmentation graph – it ideally should a little like a bell-curve
- Watch out for new unit developments or new house and land releases. They will boost medians short-term and can often have the reverse effect long term.
- Expand your sample from suburb to postcode if required. If sales volumes still too low, you may have to expand to LGA.
Kent Lardner