Tue, 22/10/2013 - 16:01
“The US house crash was evidence enough for me that the house market still lacked sufficient information. People weren’t able to see that a housing bubble had formed until it was too late,” says Allan Weiss, co-founder of the Case Shiller Weiss indexes – a series of 5,000 single-family house price indexes for the US market.
Weiss (pictured) co-founded the indexes in the 1990s with professors Robert Shiller and Karl Case and served as chief executive of Case Shiller Weiss for 10 years before it was sold to Fiserve in 2002. Commonly used by the major ratings agencies and a plethora of US banks, hedge fund managers and the wider asset management community, Standard & Poors paid to have their brand attached to the 20 metropolitan area indexes produced by Case Shiller, which are now referred to collectively as the S&P Case-Shiller index. This April, the firm was sold to CoreLogic.
Aside from the metropolitan indexes, Case Shiller produces 5,000 indexes at the five-digit zip code level. This has been the accepted norm. Until now.
The 5,000 indexes at the zip-code level give hedge fund managers, banks, home loan investors etc, the opportunity to keep track of the underlying collateral of each house. But the problem here is that the data lacks sufficient granularity. Within a single zip code there may be anywhere between 2,000 and 20,000 different homes. That is a huge dispersion profile and somewhat of a blunt instrument when trying to understand the deeper underlying price dynamics at the individual house level.
Speaking with Hedgeweek, Weiss confirms that he has developed a new solution to bridge this data gap. One that provides data at the individual house level and which gives a far more accurate picture of the current loan to value (LTV) ratio, in turn equipping portfolio managers with a more accurate input to their pricing models.
The result, based on 40 million data points covering 65 per cent of the US house market at the single-family house level, is a series of dynamic maps that uses colour coding to illustrate how house prices are shifting. For a specific area, say LA county, the user can view a series of dots where each data represents a single home. Dark green represents house prices that are rising one per cent each month, whilst pink to dark red indicates falling house prices. Users can input a time series using data going back 10 or more years, click play, and watch the dynamic maps unfold where each second equates to a one-month interval.
The beauty of this is that it paints an accurate picture of precisely how and where the US house market went from boom to bust, right up to the present day.
“I thought about what the information vacuum was that allowed the house market bubble to unfold. The conclusion I reached was that the information was not granular enough to give people an early warning. I set about building more granular indexes at the house level because what I presumed, which turned out to be true, was that certain houses suffered softening prices long before the entire zip code and certainly long before metropolitan areas. If you could have had a visualisation that started off all dark green and began to see areas of pink that would have been a warning to people that some kind of sea change was underway.
“It might not necessarily mean that house prices are going to crash but the risk of a sea change is higher than it was when everything in the map was dark green.
“These maps therefore allow people to interpret millions of data points as the market unfolds. How does one area compare to another? Does it look like a bubble has formed? Right now people speculate. The goal for these maps is to fill that vacuum that I believe still exists,” explains Weiss.
According to Weiss, it looks like the weakness originated in San Diego county where prices went from dark green to light green to pink and then crawled up the coast towards LA county. This was an early warning system that nobody at the time could have foreseen. Weiss compares it to navigating the oceans without weather maps.
A source at one major US hedge fund says that it is too early to draw any conclusions on the dynamic maps: “We’ve been pitched on this, but it still seems like early days and we haven’t really formed an opinion. We’re still asking questions and following up with them.”
The upshot to having access to better data is that portfolio managers could find themselves better able to enhance alpha generation by having a clearer picture of how house prices in a specific region of the US are changing. Weiss believes the ability to identify early price trends is a potential game changer:
“It will give people an early warning of one market moving differently to another. In addition, if a manager is looking to trade an RMBS pool they can get the index for each house in the RMBS pool and price the pool more accurately. They all use loan to value as one of the most important inputs into their pricing models but usually these pools only have the original value of the house (not the current value) and they are forced to treat every single house within the zip code identically.
“With the creation of these dynamic maps, we can help them to figure out a much more accurate LTV and price direction and make better judgment calls on where house prices might be one or two years down the road.”
Thu 15/01/2015 - 08:19
Mon 22/12/2014 - 06:30
Tue 22/07/2014 - 13:01
Tue 22/07/2014 - 12:06
Mon 22/12/2014 - 06:30
Fri, 01/May/2015 - 19:00
Fri, 01/May/2015 - 19:00
Fri, 01/May/2015 - 17:00
Fri, 01/May/2015 - 12:00
Fri, 01/May/2015 - 09:00
Fri, 01/May/2015 - 06:00
Fri, 01 May 2015 00:00:00 GMTAlgorithmic Trading Developer - C++, Linux, Python
Fri, 01 May 2015 00:00:00 GMTDirector - Core Java Architect - Equities
Fri, 01 May 2015 00:00:00 GMT