DebtWatch No. 44 April 2010: House Prices Are Not Normal
on April 6th, 2010 at 3:39 am
DebtWatch No. 44 April 2010
House Prices Are Not Normal
“I think it is a mistake to assume that a riskless, easy, guaranteed way to prosperity is to be leveraged into property. It isn’t going to be that easy.” (RBA Governor Glenn Stevens, Sunrise Program March 29 2010)
I applaud Glenn Stevens for making the above statement on national television. It was both courageous, and a succinct and accurate statement of the delusion that has come to dominate economic thinking in Australia. He effectively acknowledged that Australia has succumbed to a Ponzi Scheme: the belief that the entire country can make a living from unearned income. This something that, until recently, most public and private commentators have been strenuously denying. The great pity is that this realisation was so long in coming, while the farce is that one wing of Australia’s government has now declared its intention to bring down a Ponzi Scheme that the other wing is trying to maintain.
The data that led Stevens to this realisation is pretty obvious: the most recent quarter saw the largest increase in house prices since the ABS began keeping records in 1986.
The role of the Federal Government in causing this bubble–and earlier ones–via the First Home Owners Grant is also obvious. While previous manipulations of the market by the Grant turned a tepid rate of increase into a bubble, this time the Grant turned the fastest rate of fall in house prices into its greatest rate of increase. The current volatility of house prices is telling: eight of the ten biggest movements–in both directions–have occurred in the last two years.
With the RBA likely to increase rates specifically to prick this bubble, the volatility will doubtless continue. But even without the RBA’s expected–and I have to say justified–anti-bubble interest rate intervention, the real estate market is, as Stevens argued, far from a stable route to riches.
House Prices are Not Normal
One of the great fallacies of conventional “neoclassical” economics that encouraged behaviour that caused the GFC was the proposition that asset prices are “Normal”–in the sense that their volatility fits the pattern described by the “Normal Distribution”.
The superficial beauty of the Normal Distribution is that the behaviour of a variable can be reduced to just two numbers–its mean and standard deviation. But its real, deep beauty is that if a variable follows a Normal Distribution, then extreme events are vanishingly rare. if a variable moves normally, then:
- Movements of 5 standard deviations or more above or below the mean are so rare as to be effectively non-existent; and
- Their rarity means that they play no significant role in shaping the system: its behaviour is completely described by the events that fall within the +/- 5 standard deviations range.
As stock market speculators learnt to their great cost during the GFC, that is so not the case for share prices–since “impossible” or “Black Swan” movements in prices have been the order of the day since 2007.
For example, the average daily movement on the Dow Jones since 1914 is 0.028%, while the standard deviation is 1.13%. If stock market price movements were “Normal”, there would have been just one daily decline of more than 4.5 percent since 1914. In fact, there were 100 such falls, out of a total of 24,593 daily movements in the Index–fully 100 times as many falls as a Normal distribution would predict.
Nor could those falls be ignored in the long run: they caused a collective 612.5 percent fall in the Index, when the sum of all the percentage movements since 1914 is 688 percent.
Anyone who relies upon the Normal Distribution when investing in the stock market is ultimately on a hiding to nothing to lose his shirt, because the Normal Distribution seriously underestimates the odds and the importance of extreme volatility in share prices. A far better guide to how share prices actually behave is the “Power Law”, as well as Didier Sornette’s research based on an analogy to earthquakes.
So how do house prices stack up? Though we have a far shorter time series for house prices than for shares, one thing is for certain: house prices are not Normal. The mean quarterly change in the ABS series for nominal house prices since 1986 is 1.24%, and the standard deviation is 1.786%. If house prices were Normal, the distribution of quarterly changes would look like the red line in the next chart; the actual pattern is shown by the blue bars.
The vast majority of quarterly movements are below the mean, with the largest number–27 out of the 93 quarters–registering just above zero change (an average of 0.267% increase for the quarter). The high overall average of 1.24% growth per quarter in nominal house prices is driven by the smaller number of quarters (26 out of the 93) with increases above the average.
The data is skewed in time as well as magnitude. A truly Normal distribution would have no time pattern to the data, with a large movement just as likely to be followed by a small one. The actual distribution has long periods of low increases with clusters of large changes–and these have increasingly involved large falls as time has gone on. The next chart compares the actual pattern of price movements (in red) to a simulated random pattern (the black crosses).
There are several movements–especially the -3.4% and +7% recorded since the GFC began–which are outside the standard range for a Normal Distribution. They are not so far outside that we can categorically say that a Power Law accurately describes house price movements, as we can with share prices. But the odds are that these two leveraged asset classes share the same fundamental dynamics.
The FHOG of Real Estate
It should also come as no surprise that the First Home Owners Grant scheme significantly distorts the housing market. From the statistics, there is no doubt that the true beneficiaries of the scheme are vendors, real estate agents, and lenders–not first home buyers.
There are several ways to slice and dice the data on this point: there are years when there was no Grant, and years when there was a grant in some form or another; periods prior to the introduction of a Grant, or a change in its magnitude, and periods after the change; and periods when the Grant was doubled. The following charts show these dissections.
Periods without a FHOG had significantly lower growth in house prices, and significantly lower volatility in prices. The average quarterly price change without a FHOG was a mere 0.44%–one third of the average for the entire series. The volatility was also substantially lower, with all movements being between -1 and +2.5%.
Periods with a FHOG had both substantially higher average price rises (2% p.a. vs 1.25% for the entire set) and substantially greater volatility (ranging from -3.4% to + 7%).
A closer look at the impact of the FHOG shows that its role is that of a storm trooper for the housing market. The next chart looks at the movements in house prices in the 2 years after an introduction or change to the Scheme, and in particular at what happens to prices in the 2 years after the payment was doubled (in 2001 under Howard and 2008 under Rudd). The “Pre-FHOG” is all other quarters apart from these 2 year segments.
All but one of the large increases in house prices (4% or more in a quarter) occurred after the FHOG was doubled, while the average quarterly change in prices was over 2.9 percent. If the FHOG is the real estate sector’s storm trooper, then doubling the FHOG is its Panzer division.
The next table summarises the statistical properties of house price changes, including “Kurtosis”–a measure of how peaked the distribution is compared to the Normal Distribution–and “Skew”–a measure of how biased the distribution is towards above or below mean movements. Periods without a FHOG have a peaked distribution (Kurtosis greater than zero) and few price changes below this peak with many above (Skew greater than zero); periods with a FHOG have a flattened distribution (Kurtosis below zero, meaning that price movements are more widely dispersed), and a negative skew (meaning that there are more price movements below the mean than above).
The role of the FHOG in causing house prices to rise faster than consumer prices is even more apparent if we consider the annual CPI-deflated series–but what is then also obvious is its decreasing effectiveness over time. When rolling annual price changes are considered–a more realistic time frame for changes in house prices, since this is a slow moving asset market–the biggest price inflation bang for the FHOG buck was back in 1988, when the rate of increase hit almost 30%. Howard’s doubling could only score a 16.5% maximum rate of growth of real house prices; Rudd’s has thus far peaked at 11.25%–though this omits the impact of the most recent 7% increase in nominal house prices (since CPI numbers are only available till December 2010).
The real house price data emphasises the message that the real beneficiaries of this government intervention are not first home buyers, but vendors, real estate agents, and banks–in increasing order of benefit.
The vendors benefit from a higher price; the agents benefit from higher turnover and fees; while the banks benefit from the increased mortgage debt that first home buyers–and then the vendors they sell to–take on in order to buy into a government-supported Ponzi Scheme. The banks and mortgage lenders in general have been the biggest beneficiaries as mortgage debt has risen from under 20% of GDP in 1990 to over 85% at the end of 2009.
The revival of this Ponzi Scheme played a key role in Australia’s sidestep of the GFC. As is obvious in the next chart, the mortgage debt to GDP ratio began to fall prior to the First Home Vendors Boost, but then accelerated once the Boost was available.
The Australian economy has thus returned to debt-driven growth, with the household sector carrying the full burden for the private sector. I remain sceptical this period of debt-driven growth will last as long as in previous bubbles when our private debt to GDP ratio was half what it is today.
Table One
Table Two
End of Debtwatch Report
I started Debtwatch to raise the alarm about the approaching financial crisis that we now call the GFC, and to raise awareness of the unconventional economic theories of Hyman Minsky. Forty four Debtwatch Reports later, those objectives have been met, and maintaining the pace of one major report each month is now hampering my ability to write a book length treatment of the Financial Instability Hypothesis.
I signed a contract for this book with Edward Elgar Publishers back in 1999, with the intention of delivering it in 2001. Then I decided to write Debunking Economics, which I thought would take just six months. 18 months later, I finished it, and the debate it caused with neoclassical economists (due to my novel critique of the theory of competitive markets) took up another 4 years.
I planned to start Finance and Economic Breakdown in January 2006—and in December 2005 raising the alarm about the GFC took over.
Now I really have to give the book first priority. With each Debtwatch Report taking something close to a week to write, I can’t do both. So I am going to cease publishing Debtwatch.
What I will trial instead is publishing a monthly update on the book. I will no longer send this out as a PDF, but will make it my monthly blog entry.
bb,
Thank you for your valuable contribution to the discussion! It is good that arguments against the presence of the housing bubble are presented in so clear and professional way.
I may agree that “prices are inelastic in the short run (1-2 years) but elastic in the long run. It takes time between seeing real momentum in prices and delivery of new stock. Local council approvals, re-zoning, and infrastructure can take years to deliver – but it will get delivered.”
I would say that this says more about the shifting supply curve in outer suburs. You have clearly identified the lag in the process of producing additional supply of houses – between the time pricing signal appears and finally houses are sold. This lag may be crucial in studying the dynamics of the system. I would add that there is a clear trend of rising nominal land prices (maybe a kind of “ratcheting”) superimposed with the dynamic fluctuations but that trend only plays role when we look at the market in time scale of decades not years. In that time scale the CPI inflation and the shift in general structure of prices must also be included in the study.
I was trying to get some data to describe the supply side for outer suburbs of Sydney however this was not easy. I believe that a standard 500sqm block of land may have cost $200k in 2002 and now costs about $300-330k in the area where I live (Blacktown/ Quakers Hill / Stanhope Gardens).
The costs of construction may not have risen significantly since I arrived in Australia in 2003. When I was doing an informal valuation of my house for insurance purposes I assumed more than $750/sqm but this was not an assessment based on scientific basis. Prices may not be a linear function of size for small houses.
We may come to a conclusion that new house prices and as a consequence house prices in general in outer suburbs are anchored quite firmly to the dynamic equilibrium of the supply of new housing stock and demand.
However I would argue that:
1. Apartment and house prices in inner suburbs may not be anchored firmly in the costs of supplying new stock because the supply may be quite limited.
2. The critical component determining house prices is the dynamics on the demand side.
I don’t want to hide my intention of (getting someone to) build a dynamic model of the market to verify the hypotheses listed below. Or doing it by myself but I may have not enough time…
Please have a look at this:
http://www.systemdynamics.org/conferences/1992/proceed/pdfs/hu247.pdf
http://www.systemdynamics.org/conferences/2009/paper_thread.html
(check out the “Economic Dynamics” section)
I would add that independent variables in the model should probably be (I have not ordered the factors by the severity of impact):
1. Population growth rate and the social profile of growth (whether this is accomplished by skilled migration or other factors)
2. Income growth (related to GDP changes and changes in taxation)
3. Unemployment rate
4. Changing expectations in regards to house size (this may be difficult to quantify and may not be an independent variable at all)
5. Interest rates
6. The structure of taxation on land value, capital gains and factors related to negative gearing.
Regarding the demand side at least the following needs to be investigated:
A. Is the demand elastic as a function of costs of acquiring financing?
B. To what extent the demand is determined by migration?
C. How does the size of houses change as a function of other parameters?
D. To what extent the spatial distrtibution of demand depends on independent variables?
The purpose of building the model is to confirm one of the following:
Hypothesis A. The demand for housing assets mostly depends on external factors and does not significantely depend on past price flustuations (determining expectation in regards to future capital gains). External factors may have been listed above in points 1..4 or more of them may exist. THERE IS NO BUBBLE AND IN THE ABSENCE OF AN EXTERNAL SHOCK PRICES WILL NOT FALL SIGNIFICANTELY. They will saturate or slide down only slightly when the supply finally meets the demand. I believe this is what you have proposed.
Hypothesis B. The demand for housing assets depends on expected capital gains which are a function of past price rises. THERE IS A BUBBLE AND WHEN THE PRICES STOP RISING A SIGNIFICANT GROUP OF INVESTORS WILL START FIRE SELLING OF ASSETS LEADING TO A COLLAPSE OF PRICES. This is what Steve has proposed.
Note. The possible positive feedback loops (like rising unemployment leading to further bankruptcies) can be simulated outside of the actual housing sector model.
bb,
Every prospective home buyer I have spoken to in the last few years has indicated an expectation of significantly increased pricing in the future. This is rational but purely ponzi reasoning.
You mention that the government building of school canteens is misallocating resources. It must therefore be increasing the building cost for home owners.
I agree that it is hard to define a bubble and even harder to establish true intrinsic value in an asset. But I can imagine that if there is a long running asset bubble, the entire economy can be “distorted”. With wages and other costs inflated along the path.
Why is it so costly to service the englobo land?
My guess is that at every step higher wages and higher margins are being paid than was previously the case. This includes the contractors that lay the roads and the council employees. Not that these people are necessarily doing it easier than before as the cost of living has increased significantly for many due to the cost of housing.
Some of this will be reversed as the bubble disintegrates.
The depth of our deflation may be limited by the award working conditions and other forces underpinning incomes. But then again how many builders do you know who are permanent employees under an award?
All,
Does this sound familiar? (Prof Michael Pettis talking about the strategy used by the Chinese to clean up a banking crisis 10 years ago) It should we used these strategies in the late 1980s/early 1990s when Westpac was about to fail (yet again), State Bank NSW was broke, Bank SA was bust (that one was due to internal fraud BTW but no prosecutions) etc, etc.
Who will pay for China’s bad loans?
http://mpettis.com/
(My Emphasis and Edits)
“There are many ways to resolve banking crises, some more visibly and some less so – just no way to resolve them costlessly, and the key is to figure out the true cost and how it was paid. As I see it there were mainly three sets of tools Beijing used to manage the sharp increases in bad loans that threatened the banking system a decade ago, and of the three, the two most important were not explicit and so not easily measured or noticed.”
“All of these required forcing down interest rates so as to pass the bulk of the cost onto bank depositors, and so all of these had an adverse impact on the quality of Chinese growth. In other words the previous cost of the banking crisis was not a banking collapse, but that doesn’t mean the cost was easy to absorb.”
“The first of the three tools used to manage the banking crisis involved reducing the accumulation rate of NPLs, basically by keeping borrowing rates low.”
“The PBoC repressed both lending rates and deposit rates to allow struggling borrowers debt forgiveness and some breathing space. Of course households paid for this in the form of very low returns on their savings (and in Australia’s case, super funds).”
“The second of the three sets of policy tools, and the only very explicit one, involved infusing the banks with additional equity.”
“Finally and most importantly, the third way of cleaning up the banking crisis involved the central bank mandating a wide spread – probably around 1.5 to 2.5 percentage points more than the normal spread – between the bank lending and the deposit rate, which increased bank profitability substantially and so helped to recapitalize the banks.”
“In other words not only were depositors “taxed” for the clean-up by having to fund the very low lending rates, but they were taxed a second time to guarantee sufficient bank profitability to rebuild capital.”
“This in effect represented a large transfer of income from the household sector to the banks, to government and to businesses, equal annually to several percentage points (of GDP).”
I will add a fourth that we practiced in Australia in past bank crises. Once the banks have expanded their margins at the expense of savers and politically selected classes of borrowers, obtained more capital from share issues and hybrid debt-equity securities and lowered their borrowing costs THEN THEY START RAISING INTEREST RATES. As marginal borrowers are foreclosed the cost of the loss of principal is shared between the bank, the Govt (= taxpayers), the foreclosed borrower and savers.
You can only institute the 4th step when the Govt/Banks feel confident that there are enough new borrowers to pick up the RE pushed onto the market by the inevitable surge in foreclosures. Where are they going to find this fresh crop of borrowers?
I suggest the PTB in Australia are attempting to facilitate Step 4 by relaxing the FIRB rules, through the inward migration program and further incentives for RE investment. Capital wins.