Ireland’s residential property market is relatively small and heterogeneous. This creates challenges measuring house price inflation, particularly in a monthly index such as the RPPI. It leads to considerable variability in the index from month to month, variability that becomes more pronounced the more the RPPI is broken down into sub-indices. Therefore, the actual price change in any given month may not reflect the underlying price trend.
Most analysts of the residential property market are interested in the underlying trend, not necessarily the latest monthly increase or decrease. For this reason, the CSO always recommends using the 12 monthly price changes, rather than just the latest monthly price change, as the key measure of house price inflation.
To better reflect both the trend and the latest price developments, the CSO also applies data smoothing to the RPPI. Properly applied, data smoothing both emphasises the trend and leads to an earlier identification of market turning points. Additionally, data smoothing enables niche sub-indices (for example, regional house price indices) which would otherwise be too volatile to publish.
Data smoothing improves the quality of the RPPI, providing an accurate measure of both the short and long term developments in house price inflation and for more detailed information on specific market sectors.
When we look at the results of the RPPI before we apply data smoothing, the inflation rate of the aggregate indices will always be between their components. For example, the national inflation rate for all residential properties is always between the national inflation rate for houses and the national inflation rate for apartments. In the smoothed RPPI this is not necessarily the case.
To optimise the balance between reflecting the trend and responsiveness to the latest developments, the RPPI is first aggregated from its components. Then all the RPPI indices are independently smoothed. The degree of smoothing is tailored to the intrinsic volatility of each index, the aggregate indices being smoothed less than their component sub-indices.
Because of the tailored independent smoothing, aggregate changes in house price inflation may, on occasion, exceed the changes in all component indices. This does not signal an error in the aggregate index compilation. It merely reflects the different rates of responsiveness of the various indices to the latest available data.
For example, the graph below shows the annual inflation rates at national level for all residential properties, for houses and apartments, for the period August 2016 to July 2017. The national index for all residential properties is an aggregate of the national houses and national apartment indices.
|X-axis label||National-all residential properties||National-houses||National-apartments|
As you can see from this example, it seems that the overall inflation rate is less than the inflation rate for either houses or apartments in August 2016. It also seems that the overall inflation rate is greater than the rate for either the houses or apartments in July 2017. However these results have come about from the smoothing technique used. As explained above, smoothing improves the quality of the indices but it also causes this anomaly in the growth rates.
Details on the data smoothing technique can be found in page 10 of the RPPI Technical paper