Revisiting Steve Keen’s 40% property price fall prediction

Anybody who follows the property market in Australia is familiar with Steve Keen, an economist who made a daring prediction that our property prices will fall as much as 40% sometime before 2023. How this prediction came about is subject to various interpretations, nevertheless, the “40% drop” forecast has been repeated so many times over the years, and by so many, that it has almost become a part of Australian folklore.

I have to admit that when I first came across Steve Keen’s prediction I treated it with a fair dose of scepticism since I could not find corroborating evidence for such a substantial fall in prices using indicators from my own analytical toolkit.

I had a particular issue with the somewhat overly simplistic assumption that 'whatever happened in the US has to happen here because we are no different'. I had an alternative perspective on this as, a decade or so earlier, I had undertaken some analysis of the 1980s housing crisis in the US and clearly, the events played out quite differently in Australia.

However, there was one bit of theory that underpinned Keen’s conclusions which attracted my further interest. In particular, at the core of Keen’s argument is a concept that “aggregate demand equals GDP plus the change in debt” (emphasis mine).

Since debt is equivalent to cash in your pocket and 'money makes the world go round', it therefore seems a logical proposition to look into changes in debt levels as a way of gauging what is happening in a wider economy.

Keen devised an indicator called the Credit Accelerator to forecast the effect of changes in debt levels on GDP. It is conceptually similar to Credit Impulse proposed by Biggs, Myer, Pick in their analytical paper published in July 2009.

Briefly on Credit Impulse, as it is important for the interpretation of the graph presented below: the original theory was the result of an attempt to find a possible explanation for credit-less recoveries where GDP starts to grow with no growth in stock of credit. The authors found the evidence that a mere slowdown in the speed of debt reduction (i.e. in deleveraging) had a positive effect on the overall economy.

In simple terms, when the Credit Impulse indicator starts to point upwards while in negative territory, it is a positive sign that things are improving. The opposite should also hold true; that is, when it starts to point down while in positive territory - it is a negative sign meaning things are deteriorating.

So by substituting total debt with just housing debt, one can potentially gain insight specifically into the property market.

However, it is debatable whether Credit Impulse/ Credit Accelerator are still meaningful in this context. And there are also some issues with refinancing not being able to be isolated from the total value of housing debt which potentially clouds the picture. Nevertheless, a comparison of Credit Impulse (constructed with housing debt data) to the 12-month rate of change in the established house price index indicates quite a good correlation between the two data series, as illustrated on the following chart.

Source: Based on ABS and RBA data

I am using an adaptation of the Biggs, Myer, Pick version of the algorithm, hence there may be discrepancies between Keen's and my version of the graph.

It is worth noting that correlation between the two data series is particularly strong over the last five years (for those statistically inclined, correlation coefficient is 0.88 but drops to 0.64 for the entire series). That would imply there is merit in using the Credit Impulse concept for monitoring the state of the property market in Australia.

The potential benefit of this indicator is twofold: it offers a timely insight into the dynamics of the property market – better than quarterly established house price data from the ABS. And secondly, perhaps more importantly, it can provide an early indication of impending changes in the direction of property prices.

As the above graph demonstrates, the theory behind Steve Keen’s original prediction appears sound - although it is not entirely clear to me whether Keen had actually been using his Credit Accelerator indicator when he made his prediction or not (Biggs, Myer, Pick paper was published much later). However, let’s examine how this indicator should have been interpreted if it was in existence.

Article continues on next page. Please click below.

To be frank, it would have been a frightening experience to stare at this graph in 2007 and 2008, hearing about all the financial collapses happening throughout the world and not knowing the depth it could all fall to.

Based on these factors alone, a prediction of significant declines in house prices would have been well justified back then. However, since the magnitude of the likely decline cannot be derived directly from the Credit Impulse indicator, I assume there must have been other factors that allowed Keen to make the estimate of 40% drop in prices, because otherwise it would have been just a presumption.

By the same token, when the Credit Impulse started to point upwards in March 2009, and kept rising in the following quarters, it gave an indication that things were likely to improve. However, this message was not conveyed to the wider audience. So, again, one can only assume that there must have been other compelling factors that led Keen to persist with his earlier conclusion.

The whole situation repeated again in 2010 (i.e. in June Credit Impulse indicated a possible reversal in the growth of property prices) and in 2012 (when the Credit Impulse started to point upwards again, indicating the likely end to price decreases).

By then Keen had been regularly publishing commentary on his Credit Accelerator indicator, so the experience from the earlier 2008-2009 episode should have helped with a more accurate interpretation of what was happening. However, throughout that time, Keen had insisted that there were only two likely options for property prices: a US-like scenario where prices fall rapidly, or Japan-like scenario of a “slow melt” over the next decade or so.

Again, this call was made most likely because of other factors which, to my best knowledge, remain unspecified (that is, if we leave aside the two simplistic assumptions Steve Keen made that 'Australian private debt is high and therefore it will deleverage' and that 'Australia is like the US so the same has to happen here as well').

It was only much later, when a third scenario started to play out and property prices actually recorded some gains, that Steve conceded prices “may go up for a while”. However, the media did not pick up on this comment with the same zeal as on the previous doom scenario. How unfortunate.

All in all, in 2008 Keen had an important message to tell and that message was broadcast to a wide audience. Media played a very important role in publicising his warning. That is how I and many others became aware of his views.

However, not many took the effort to understand what was really behind the “40% price drop” headline. The majority bought into this scenario without any further thought.

Many still hold to this view despite prices reaching new heights and despite the fact that the indicator that is at the core of the original prediction is showing that property prices will grow further. However, the positive message is not being reported by the media.

To be fair, there is a small possibility that the “40% price drop” prediction may come true in the next decade (the timeframe nominated by Steve Keen). We can only wait and see. However, if it happens, without additional evidence from Steve on what kept him convinced that the original scenario is still to play out, it will be difficult to accept that it was all predictable as far back as 2008.

One thing is almost certain though. Based on the past performance, the Credit Impulse should give a warning when the current price trend is about to reverse. For now, let’s just enjoy the ride to the next top, however long, or short, it will be.

Arek Drozda is an independent analyst who has worked in the public and private sectors for over 20 years in business development, data analysis and in building geographic information systems.