The RP Data-Rismark Home Value Index results were released earlier this week revealing a 1 per cent jump in dwelling values across Australia’s combined capital cities. The rise comes on the back of a -1.4 per cent fall in May and brings the cumulative decline in home values to -1.2 per cent over the June quarter and the first half of 2012. Capital city dwelling values have fallen by 6.5 per cent since their peak back in October 2010.
The June rise of 1 per cent has thrust the index methodology back into the spotlight with one commentator in particular making incorrect claims about the indices accuracy and reliability. Based on these comments and the media attention they have attracted it is probably timely to provide a refresh on how RP Data, together with Rismark International compile the daily Index and address some of the miscommunications that have been given some airplay over the week.
As a first point, it is important to note that the RP Data-Rismark Index is the only private measure of housing market conditions that has been independently peer reviewed (you can see the results here). Additionally, the methodology is completely transparent with several white papers available describing the technical detail behind the index calculation. It is fair to say that these papers are a hard read for someone without a high level statistical knowledge; this index is a sophisticated measure that uses complex mathematical formulas to produce the results. The description of how the indices are produced is inherently complex.
It is hard to take the assaults on the Index credibility seriously when it has the support of Government and private agencies that are well qualified to assess the reliability and accuracy of the results. As you can see from the official chart pack delivered by the Reserve Bank of Australia this week, the RP Data-Rismark Home Value Index is the only measure of housing market conditions referred to. Additionally the Index is widely quoted by Australia’s leading economic commentators much more so than other index measures.
It is worthwhile noting that the RP Data-Rismark Index shows a 98.6% correlation with the annual returns reported by the ABS House Price Index which is published quarterly. Of course, over shorter periods, an index which uses information about the locations and attributes of the properties which have sold will capture shorter term market movements which are missed by other methods.
While we agree this is a reliable and useful index it suffers in timeliness (the June results won’t be available until August 1), excludes sales other than detached houses, and doesn’t have the granularity of a monthly (or for that matter, daily) measure of market conditions. It will be interesting to take note of the June results from the ABS when they are released; my bet is that the outcome will be remarkably similar to the RP Data-Rismark result which showed a 1.2% fall over the three months ending June.
Why are RP Data and Rismark computing a daily index?
The Australian housing market is worth an estimated $4 trillion; that’s three times larger than the value of all stocks listed on the ASX and three times larger than all superannuation assets in the country. This is Australia’s largest asset class by far and it is important that we have the most sophisticated measure available to understand how values are changing from period to period. The daily measure, for the first time ever, provides insights about the intra month value movements and a greater understanding of the risk and volatility of the housing market; home values do not move smoothly from month to month, there is market volatility.
The daily reporting of the index also opens up opportunities for financial markets allowing people to gain or hedge exposure to Australia’s largest asset class. The daily provision of the index is an essential component is establishing a tradeable mechanism. The daily index was designed specifically to track the risk and return profile of a diversified property portfolio and enable financial markets to use the index as benchmark for trading. The ASX has selected the RP Data-Rismark Index, after completing their due diligence on which index and index provider is the best placed, for the basis of trading in Australia.
Below are some of the typical questions and answers that we have been fielding since the daily index was released (a special thankyou to Dr. Matthew Hardman who heads up Rismark’s research team and was the principal architect of the daily index methodology).
How can the small number of sales on one day tell you how much the whole market has changed?
By themselves, they don’t. We don’t throw away the previous day’s sales. A house that sold yesterday or last week has almost as much relevant information as a house that sold today. The time lag of previous sales is taken into account statistically via the hedonic imputation method. It gives us the best estimate of the value of our market portfolio, based on all the available historic information.
But you don’t know about every sale as soon as it occurs. While some sale prices are reported the day of the transaction, others take weeks to arrive in the database. Does this make the index less accurate?
We record all sales and use them as the data arrives in our database. We use statistical methods to account for the timing difference between the sale date and the date we receive the information. If we receive information about a sale one week after it has occurred, we use that sale to update our current view of the market, mathematically taking into account that the sale occurred one week ago. The information in that sale is still relevant to the current state of the market and our model incorporates it.
All indices must have a mathematical method for dealing with the sales data which arrive after the actual sale date. The hedonic imputation method is the most effective in doing this.
What if interest rates rise suddenly? This should dampen property prices. How will the index account for events like this?
The index will react when it receives sales information showing sufficient numbers of sale prices higher or lower than those expected by the previous day’s model (which did not know about the sales). It cannot react until given this information because it will have no evidence to cause it to.
If the index has been falling for the last several months, but has risen over the past few days or weeks (or vice versa), does this mean the market is turning around?
No more or less than if it happened in the stock market.
All financial market price series have natural volatility. Prices can fluctuate over shorter periods, while showing a longer term trend up or down.
If you want to gain a sense of whether or not the momentum in a property market really is changing, it is important to look at the trend over longer time intervals, such as the last quarter or last six months and filter out the short term noise or volatility which occurs in all markets.
Is the downward / upward trend starting to flatten? Is the flat trend starting to turn up / down?
Will the index be impacted if only less expensive properties sell over a given period? Will it show a fall in the market?
Preventing the index being “tricked” like this is one of the main strengths of the hedonic method.
It is important for the index to distinguish between
- Lower observed sale prices due to a larger number of lower value properties selling.
- Lower than expected sale prices of those particular properties which have sold.
Only the latter is demonstrable evidence of a downward market movement.
Indices which do not control for the attributes and locations of the properties which are observed to sell and which cannot compare an expected sale price with an actual one are much more likely to either over or underreact to market movements, because they cannot distinguish compositional bias in the sales sample from genuine market movements.
Some indices revise as more sales data comes in. Why doesn’t this one?
The indices give the best estimate of the value of a diversified investment portfolio, using all the information available at the time of publication. Each day this information is updated on average by 1,400 transactions which in turn inform the next day’s best estimate. From testing, errors in the index are expected to be only a few basis points.
Indices which only use transactions which occur during the measurement month (or quarter) and revise as additional data pertaining to the particular month (or quarter) arrives, never have an ability to convey the current state of the market. Not only do buyers and sellers of individual homes want to understand the best estimate of the current state of the market, it is essential for market participants wishing to settle contracts over the market portfolio.