Buying a House: Suburb Selection Methodology

I cannot even start to use the search features on the real estate aggregation sites until I pick a suburb or two. This post captures my ad hoc thoughts on suburb selection and narrows down the list to specific suburbs that meet my acceptance criteria for buying a house.

What is a good suburb to look for a house to buy?

To start with, I figured it might be useful to narrow the search to the East and North train lines and list all suburbs that fall within zone one. I looked at the metro maps for the lines in question, then surveyed Google maps by hand and wrote down the lists of candidate suburbs, a crude but functional starting point:

• East (Belgrave/Lilydale Line): Richmond, Cremorne, Burnley, Hawthorn, Hawthorn East, Camberwell, Canterbury, Surrey Hills, Mont Albert
• South East (Alamein Line): Richmond, Cremorne, Burnley, Hawthorn, Hawthorn East, Camberwell, Burwood, Glen Iris, Ashburton, Ashwood
• South East (Glen Waverley Line): Richmond, Cremorne, Burnley, Toorak, Camberwell, Glen Iris, Malvern East, Ashburton, Chadstone
• North-East (Hurstbridge Line): East Melbourne, Collingwood, Abbotsford, Clifton Hill, Fairfield, Alphington, Ivanhoe, Eaglemont, Heidelberg
• North (Epping Line): East Melbourne, Collingwood, Abbotsford, Clifton Hill, Fitzroy North, Northcote, Thornbury, Preston, Reservoir
• North (Upfield Line): North Melbourne, Parkville, Kensington, Flemington, Royal Park, Brunswick, Brunswick East, Coburg, Coburg North, Pasco Vale South, Fawkner

There are some biases built in here, specifically my wife and I are familiar with the east and south east parts of inner suburbia, and previously unsuccessful house hunting efforts have focused on these areas. This (irrational?) reduction excludes at least the south (Pakenham Line), North West (Craigieburn Line) and West (Werribee and Williamstown Lines). There may be good opportunities in those directions, but scope reduction has to come in sometime and for some reason...even arbitrarily.

Why is crunching house prices for whole suburbs a bad idead?

• Obviously, there are many many more facets to a suburb than access to a train line and the median house price.
• In this case, the suburb names were chosen by me by hand and the dataset may have holes.
• Suburbs are different sizes and have different qualities. For examples, on the East train line there are hubs of major shops at (at least) Richmond, Hawthorn, Camberwell, and Boxhill, suggesting price spikes in desire and in turn price at close to these suburbs (more accurately at or around the shops and train stations in these suburbs). 
• What is a "house price" and how is it different to a unit price? For example, where do town houses fit in (likely under units, although a town house or a house would meet the acceptance criteria)? 

These, and other issues I can't even think of off the top of my head highlight the complexity in interpreting any price data by suburb.

A likely more accurate methodology than comparing mean or median suburb house prices would be to sample asking or clearance house prices within the limit (15 minutes walking distance) of each station of interest and take a central tendency. This more time consuming approach might be better suited once the list of candidate suburbs has been drastically reduce to a handful.

Now that I have discussed what is wrong with comparing suburb prices and what could be done better, the following series of graphs provide suburb comparisons based on median house price up to the end of 2010. The data was crunched by hand from the APM House Price Guide that allows one to type in a suburb name or post code and get details of house sales in that suburb.

As with the Melbourne house price data, I used Google Docs to create the graphs. I have provided a few qualitative comments for each graph, but I suggest that none of theses summaries be used as a primary source for decision making. I intend to use it more as a tool to provoke questions and have conversations about specific suburbs.

The following graph summarizes median house price for the Belgrave/Lilydale Line. The lowest median price in these suburbs are scarcely below $750K. The data suggests a dip in the industrial inner city suburbs of Richmond to Burnley, although my "experience" suggests at least Richmond and Cremorne are growth suburbs given their proximity to the city. The glut in for the middle suburbs does not surprise me. I've added Box Hill just for fun.

The following graph summarizes median house price for the Alamein Line. This line is an offshoot of the Belgrave/Lilydale line, sharing 6 suburbs. Burwood and Ashwood are relatively large suburbs that come within close proximity of the train line, although are generally "believed" to be difficult to access via public transport.

The following graph summarizes median house price for the Glen Waverley Line. As with the Belgrave/Lilydale line, the middle suburbs on this line are desirable, reflected in their high median house price. Chadstone might be an interesting point of investigation with a median of $636K.

The following graph summarizes median house price for the Hurstbridge Line. I believe the Hurstbridge line was a great pick, but over the last 10 years or so that I have being paying attention it appears that the value there has been exploited. Collingwood, Abbotsford and Fairfield may still be interesting opportunities.

The following graph summarizes median house price for the Epping Line. I had identified the Epping line as an area of interest this time a year or two ago. The line shares some suburbs with the Hurstbridge Line, although there may be some opportunities in Thornbury, Preston, and Reservoir. The Preston market right in the middle of those suburbs helps to make a stronger case.

The following graph summarizes median house price for the Upfield Line. I have a very limited experience with this line. Royal Park is an outlier, the suburb is filled with parkland. The data suggests a look at Kensington, as well as Coburg and Coburg North may be worth a deeper look.

Graphing the data is interesting, but as I suggested, I don't think anything specific can be learned from it. My weak qualitative comments have served to reinforce/rationalize some pre-existing biases and thoughts.

Nevertheless, in summary, some suburbs that I will perform some more detailed research on include: Chadstone, Thornbury, Preston, Reservoir, Kensington, Coburg and Coburg North. For now, I'll add Surrey Hills to this list, as it was another suburb we had previously been sniffing around.

Some questions that I may seek to explore, include:

• What is a regression of asking prices of houses that meet my acceptance criteria in chosen suburbs?
• What is the estimated travel time by mass transit (train, maybe tram) to the CBD for selected suburbs?