Johannesburg - Every house has a price. But what that price is, and how it's calculated, can be debated. One thing is for sure, and that is that the different house price indices in South Africa are bewildering.
There is a place for each instrument to determines house prices, but every index has its pros and cons, says property valuer and economist Erwin Rode of Rode & Associates.
Four house price indices were announced last week, each with different growth figures. Standard Bank's numbers show a 1.8% decline on an annual basis in the median house price; Absa indicates an average nominal house price growth of 1.7% on an annual basis; FNB produces an average house price growth of 2.3% on an annual basis; and ooba shows that the average house price has fallen by 5.5% on an annual basis.
Then there is the quarterly Rode property index.
The differing figures raise questions as to the reliability of the data.
Rode says the most important factors in compiling a house price index is how representative it is of the market segments that you want to follow, and what chance it has of being distorted by a shift in the price range - for example, if at some stage sales of cheaper houses exceed sales of more expensive homes, or the converse.
"The biggest problem with the indices is how to value the same house, over time, cost-effectively." To do so with 100% accuracy would be extremely expensive.
He says that it is easy to measure the price of a tin of beans over a period, but the problem with a house is that every time it's a different house, with distinct characteristics, that is sold.
"The fundamental problem in compiling a housing index is how to overcome this."
Options
He says the best and most expensive option is to build multiple- regression models, which would be expensive to maintain.
This would involve separately researching, for instance, the quality, area and orientation of every house sold. "In theory this would be the best model, but in practice it is simply too expensive," he adds.
An alternative would be to simply monitor over time the median house price in homogeneous areas. "This method largely solves the problem of comparing dissimilar things over time, and it is cheaper."
Another problem is the issue of delays, especially when using deeds office data. The use of data from banks' own loan books results in fewer delays than the use of indices based on deeds office data.
"Those indices using deeds office data, however, have the advantage of possibly being more representative."
No ideal
He said there is thus no ideal index, because the market is so heterogeneous, but he reckons that Absa's house price index makes a valuable contribution because the bank has a large slice of the market - between 30% and 33%. "The fact that Absa has divided the middle segment into categories is a step forward."
Absa's residential property database made its appearance in 1981, out of the former United Building Society, and national and regional house price trends were first published in 1984. Jacques du Toit, senior analyst Absa's home-loan division, says the database has been expanded and refined over the years to provide residential property market data that is consistent and reliable over time.
"The information in the database goes back to 1966, and is derived from information in mortgage applications that Absa has approved."
Absa's first monthly house price index was published in May 2002.
FNB's house price index, which saw the light of day last week, is compiled using the average value of housing transactions of up to R10m that have been financed by the bank.
Standard Bank calculates the median, which is the price in the middle, where half of the houses are cheaper and the other half more expensive - in other words, the whole spectrum of houses sold. Ooba also uses its own data, rather than using general industry data.
The Rode house price index was developed in 1984 by Rode and, together with the Absa house price index, is the oldest house price index in South Africa.
To calculate price trends, the Rode index every quarter tracks house sales in a select number of suburbs, reflecting high, medium and low price categories in the most important of these.
The indices' data go back to 1972. Medians, rather than arithmetic averages, are used to reduce the effect of breakout prices on the "averages".
The Rode index is calculated on sales registered at the respective deeds offices. This means they represent the entire market but, because the index is based on sales that first need to be processed by the deeds offices, the delay is about six months. "As a result, the Rode indices are more for research than publicity," says Rode.
Britain has at least 10 house price indices which also cause confusion, just as in South Africa. "They use the same methods of calculation, experience the same differences between test samples, and differ in the results produced as well," says Du Toit.
But all the indices are useful as broad indicators of what is happening in the housing market.
- Sake 24