Search This Blog

Thursday, January 14, 2016

Role of Market Supply During Housing Crash

Many people ask me over and over whether stock on market, unemployment, or vacancy rates need to spike before house prices start falling. That thinking seems to be valid but only if someone assumes that supply side is what drives prices up during boom times. I already covered this issue before but I want to be more clear.

We can look into three indicators to get better understanding of this issue, stock on market (number of homes for sale), rental vacancy rates and number of sales. We'll also look what happened in USA because one of the main arguments for such claims are based on understanding that high foreclosure rate pushed US houses prices down.

Let's have a look in US house prices:
US house price index
 Prices peaked in late 2005, by the end of 2007 were down  20% and by end of 2008 they reached -40%. In some bubbly areas magnitudes were larger (e.g. Phoenix AZ -40% by end of 2007, -75% by the end of 2008) but timing was very similar.

The main question here is: why?
Some people say unemployment rise:

US unemployment

They are wrong. Unemployment was at multi-decade lows at the end of 2007 at what time house prices already lost half of their total losses (20% on average 40% in some cities). Even by the end of 2008 unemployment was still relatively low (under 6%) while by that time house prices fell almost to the bottom.

Some people say subprime mortgage delinquency and foreclosures:


In late 2005 when house prices peaked mortgage delinquency and foreclosures were close to the record lows while  By the end of 2007 the spiked to levels not much different to historic lows. By the end of 2008 reached levels 30% over 2002 levels. There is clear correlation between prices and foreclosures but not causation. It seems that foreclosures were behaving similar or slightly lagging in time behind prices. It's unlikely that foreclosures caused house price to fall since they were  as often during boom 2002 when prices jumped 10% and 2007 when prices fell by 20%.

Let's see oversupply - new home construction:


Construction was rising and it peaked in late 2005 at the same time as prices peaked. It fell fast afterwards and by end of 2007 was down 60%. Falling construction clearly didn't cause prices to fall during the bust. Maybe construction was so great during boom years with huge unsold inventories that later forced prices down. So, let's see if there was an oversupply of all homes - stock on market:



Number of homes for sale was rising with house prices during bubble period. It peaked in 2006 at the same time house prices peaked but than it fell with house prices. It is hard to argue that falling stock on market was driving price drops. Stock on market was above 'normal' and rising during 2000-2006 period when prices were going up. During house price collapse period 2007-2009 stock on market fell by more than 50% - clearly not driving prices down. If there was a large number of unsold new homes prior to bust they would have appear on market afterwards. The same is case with foreclosures. Despite all of this stock on market was falling with house prices.

Some people say that oversupply went into rental market:


Rental vacancy rate peaked in early 2004 when prices were booming and reached a decade low at the end of 2007 when prices already lost 20%. It's hard to imply that oversupply of homes ended up on rental  or that falling rental vacancies pushed prices down.

So if it's not unemployment forcing people to sell, nor foreclosures, nor oversupply (new construction, rising vacancy rates or rising stock on market). It's clearly not increased supply so what is it?

We should look into demand side. House price growth in years prior to 2006 was driven by price speculation, prices were rising because they were rising. Property investors were ready to pay more because they expected further growth and capital gains, occupiers were speculating that prices will be going up so it's better to jump in at any price or miss out forever. Demand was booming and it was supported by easy credit.

Number of new sales:


Number of new home sales peaked in early 2005 before house prices peaked and fell quickly to levels not seen in decades. Demand for new homes collapsed prior do price falls. How about existing homes?



Let see how it all aligns together:



In late 2004 sales peaked while prices and number of homes for sale continued rising. Around 6 months later prices peaked while stock on market continued to rise. Prices started falling at slow rate. In late 2006 stock on market started falling while number of sales continued down - and prices started plummeting despite fall in supply. Falling demand (fall in number of sales) is the only indicator leading house price fall.


12% vacancy rate in USA is not comparable to our vacancy rate because they count all rental homes that are vacant while dodgy sqm only counts those listed on one or two websites for more than 3 weeks continuously. In addition rental market in USA operates differently, largest share of rental homes is owned by businesses which makes normal vacancy rate high - over 7.5% (currently USA is experiencing rental crisis with rents skyrocketing while rental vacancy rate is over 7%)  In 2006-2008 period vacancy rate was just slightly over at around 9.5% and it peaked at 11% well after house prices collapsed in 2009.

Stock on market at the peak was significantly lower (0.5% of all homes) compared to current and historical numbers in Australia (~3% of all homes). Many of sellers in Australia are trying to get good price and they will not sell for less at the moment.  But when demand collapses  there is always enough supply (people who need or want to sell regardless of market conditions) to drive prices down on low volumes (US home prices collapsed in 2006-2009 period on record low volumes not seen since WWII). It only takes a few sellers willing or forced to accept anything to push prices into the ground.

Sydney sale volumes peaked (current cycle) in mid 20014, Melbourne in mid 2015, nationally it is probably peaking around this time, ...