A change in the method of seasonal adjustment of monthly property transaction counts will be implemented in July 2011. This moves the method of seasonal adjustment from the spreadsheet based Multiplicative Decomposition (MD) method to the method used by the X-12ARIMA program. The monthly seasonally adjusted transaction counts to June 2011 will be the first set of figures published on the new basis. All past seasonally adjusted figures to April 2005, the first month of the series, will be estimated using the X12ARIMA program.
The MD method was initially introduced as the UK property transaction series was too short to use X12ARIMA. Our analysis shows the move to X-12 ARIMA has changed the residential seasonally adjusted series by an average of around 3 per cent per month, compared to MD, with a largest difference in any month of 11 per cent in the published figures. The non-residential seasonally adjusted series has changed by an average of 4 per cent per month with a largest monthly difference of 20 per cent. However, two thirds of the non-residential seasonally adjusted counts show no change at all.
In common with normal practice with X-12 ARIMA, the optimal seasonal adjustment parameters will be determined each year at an annual review and then used without change for a twelve month period. This is designed to limit changes in the seasonal factors from month to month and therefore minimise continual revisions to the historic seasonally adjusted data.
X-12 ARIMA is a widely employed method for seasonal adjustment developed by the US Census Bureau. Although the basis of the adjustment is the same as the MD method, the X12ARIMA method introduces important extra features. Employing the X-12 ARIMA method has the additional advantage of making the method of seasonal adjustment consistent with the method used for producing many other national statistics.
For more details on the methodology of X-12 ARIMA seasonal adjustment please follow the links below.
New Capabilities and Methods of the X-12-ARIMA Seasonal Adjustment Program (PDF 1MB) (Opens new window) by David F Findley, Brian C Monsell, William R Bell, Mark C Otto, and Bor-Chung Chen Journal of Business and Economic Statistics, Vol 16 (1998), Number 2