Churn baby, churn

Last week, Delusional Economics posted an article entitled The market needs churn, which showed that the rate of issuance of new loans is a leading indicator for house prices. A follow-up article was posted by Delusional Economics the next day, confirming their findings (see The secret to house price rises).

Delusional’s posts inspired me to conduct research of my own on the relationship between changes in home prices and transactions volumes. My data sources for this exercise are as follows:

For each mainland capital city and nationally, two charts are presented below: one showing quarterly changes in house prices and transaction volumes since 2002 (derived from ABS data); and one showing annual changes since 1992, whereby home prices have been derived from the ABS and home sales from APM and weighted by population.


First, consider the quarterly data from the ABS:

As you can see, there appears to be a strong correlation between changes in the number of house transfers and changes in home prices, with house transfers tending to lead prices.

Now for the annual data derived from the ABS (house prices) and APM (home sales weighted by population):

Again, there is a strong correlation, although it is unclear which variable is the leading indicator.

Now for the capital city data.


First, the quarterly data:

Again, there’s a strong correlation, with house transfers seemingly leading house prices.

Now the annual data:

Once again, there is a strong correlation, although it is unclear which variable is leading which.


First, the quarterly data:

There’s a reasonable correlation for Melbourne, but not as strong as with the previous case studies.

Now the annual data:

That’s more like it. The correlation is strong, but it is unclear which variable is leading.


First, the quarterly data:

Now that’s a strong correlation, with house transfers also clearly leading house prices.

Now the annual data:

Again, a strong correlation, but with no indicator appearing to lead the other.


First, the quarterly data:

Like Brisbane, there is a strong correlation, with transfers leading house prices.

And the annual data:

Again, a decent correlation particularly over the 2000s, with house sales appearing to lead prices.


First, the quarterly data:

A nice correlation, with house transfers seeming to lead prices.

Now the annual data.

Again, a strong correlation, with home sales appearing to lead prices, particularly over the 2000s.


As expected, the data shows a strong correlation between changes in house transactions (sales) and changes in house prices, thereby supporting Delusional Economic’s findings. Further, the quarterly ABS data tends to show that house transfers leads house prices, whereas the annual ABS and APM data is less conclusive, with both variables tending to move in unisen.

Later in the week, I will seek to add changes in finance commitments to the above charts. Stay tuned…

Cheers Leith

[email protected]

Unconventional Economist


  1. Leith – great one. I hope you are able to expand on this at some stage, in comparing the transfer and dollar volumes of the abnormal bubble markets of Australia, with the normal affordable markets of Texas.

    the “speculatioon factor” of bubble markets such as those of Australia and New Zealand, is amazing.

    • The_Mainlander

      Nice call Hugh!


      Hope your well in NZ and not to shaken or stirred!

      Kia Kaha.

  2. ceteris paribus

    Correlations high- with sales a leading indicator by a wisker.

    How much causation between the two variables and how much different manifestations of the same phenomenon driven by other deeper underlying causes?

  3. Sandgroper Sceptic

    Great analysis – and with house transfers now down in most markets (all markets?) the future bust seems baked in.

    What could save this sinking Titanic?
    (1) Interest rates – no – IMHO the RBA will use interest rate decreases as a buffer when things really start hurting but I don’t see that as a circuit breaker.
    2) Population changes are unlikely to affect the broad trend unless we import several million Japanese at once.
    (3) Tax changes. This is a possibility,if CGT was eliminated for example that might encourage a wave of new speculation. However, with the federal budget now tight tax changes are likely to worsen the situation, ie removal of negative gearing, increasing land tax etc.
    (4) FHB changes. These bullets have been fired. Even a doubling of the FHB package or a tripling would have only a marginal effect as most people who want to buy have probably already done so.
    (5) Something I have missed?

  4. Just thought I’d point out that the first of your Adelaide graphs is actually the Melbourne graph.

    • Torchwood1979

      Dublin, sorry Melbourne, is a miracle-turned-train wreck waiting to happen IMO. And I have’t read the article. I doubt I could stomach much more “Interest rates to blame!” nonsense. It couldn’t be that housing is too expensive, could it?

      • The_Mainlander

        Sir, the Emperor has no clothes but the ‘finest see through clothes’ in the land, and these threads are still cheap compared to Melbourne house prices.

        Fine thread!


        (I agree with you Torchwood 79 – it is just too Bloody Expensive in Melbourne and correction has started.)

  5. “Again, there is a strong correlation, although it is unclear which variable is the leading indicator.”

    I thought you guys read Steve Keen?

    It’s very clear from the Prof’s *detailed* analysis that debt leads housing prices by 9-10 months.

  6. Leith,
    As a long time follower, I see you use the major cities for your graphs/stat’s, for obvious reasons. We here in Darwin have a hyper inflated market with a small population. You won’t get a family home in the worst suburbs for under 550-600K (up from 150K 7-8 years ago). Currently we have hundreds of hi rise apartments that have sold off the plan for up to and over 900 K maybe average 575k off the plan(average household income is $70k apparently). The rental market is inflated, a 3 bedroom unit within 5km of the city will set you back 5-700 per week(25-35K per year!!!), house in the outer suburbs $600 per week. A drive into town shows a lot of darkened windows in apartment building at night (empty?). Cost of living is relatively high as well.
    While I understand high prices in cities of millions I can’t fathom it here where the population would be less than 200k.

    Now to my point any chance of including an outlier like Darwin in your analysis? Particularity on the graphs showing all capitals. I think it would offer an interesting case study.

    Thanks for your time, keep it up.

  7. Am guessing that this is a pretty good indicator of tightness of supply and that (in recent times) the price is controlled more by vendors than by buyers.

    If you have lots of houses to sell (i.e. too much supply), then you would generally expect when vendors lower prices then they sell more houses.

    As your graphs show the opposite, that would say to me that supply is restricted, and so when the price gets above a level then more vendors are prepared to sell, but that below that level the bulk of vendors are just happy to hold on and wait for higher prices.

    Overall seems to support the theory that housing prices will “hold up” rather than free-falling – that is unless we hit the point where the housing faithful lose their faith in the housing market (or ability to pay)…

  8. A good point Leith, I think you are probably right though of course the opposite could be argued.
    It is an attractive idea, but the linkages are more complex, I think, and it is possible to argue the reverse, e.g. if house prices are rising, more transactions will occur as people move more quickly to get into the market. If prices are falling, people take their time buying since any delay is likely to be profitable, so the number of transactions will fall, so perhaps there is feedback both ways.
    Take your pick. This topic is a good candidate for a multivariate analysis, where other variables, such as the extent to which the market is over or under long run averages, the pace of market expansion, etc. are also modeled.