In our example, it is evident that there is a linear (straight-line) relationship between sales price and the area of a unit. By using an appropriate software package, one can calculate the equation for the line that best fits the data. In our example this equation is:
Market Value = 2314,6*Area + 319228
One can now use this equation to estimate the market value of any unit in our hypothetical block of flats. Let’s say we want to calculate the market value of a unit with an area of 88m².
Market Value = 2314,6*88 + 319228 = R522.193 (say R522.000)
In reality, of course, there are likely to be a number of variables (or predictors) that explain the difference in value between the units in a block of flats, or between houses in a suburb. Some of the other predictors of value may be:
- Plot size
- Size of outbuildings
- Area of house
- Quality of construction materials and finishes
- Condition of overall property
- Proximity to a busy street
A CAMA model could include any combination of these variables, and other variables. The benefits of CAMA when it comes to the large-scale valuation of residential properties in a municipal district, says Louw, is that it is cost-effective and reduces the potential for human error and inconsistency that could, for instance, result where more than one valuer would be required to cover a particular suburb.
“A key CAMA benefit is that it is obviously much better and more consistent than the human mind at identifying and weighting the contribution of the individual predictors to the value of a property. This is particularly so in the case of mass or large-scale valuations, where it is imperative that equity is achieved across the board.
“In view of the fact that the new Municipal Rates Act requires the revaluation of all properties every four years, CAMA would also result in great cost savings for municipalities.”
He cautions that, as always, the garbage-in-garbage-out principle applies. Great care must, therefore, be taken that the models are well specified and that the quality of the data upon which the models are built is flawless. It is, therefore, not a good idea to use data collectors who are not experienced and well trained.
CAMA works well in approximately 85% of cases, but given the nature of regression, valuer intervention will be required in the case of heterogeneous (diverse) neighbourhoods such as Clifton in the Cape Peninsula or Sandhurst in Johannesburg. Because few houses in these neighbourhoods are similar, and few sales take place, the market tends to be “inefficient”. This means there are great differences in prices because imperfect information is available to the market. However, as Louw points out, “manual” valuation approaches encounter the same problems in heterogeneous neighbourhoods.