• Summary
• Downloads
• Explanatory Notes
• Related Information
• Past Releases

Effect of new seasonally adjusted estimates on trend estimates


TREND REVISIONS

Recent seasonally adjusted and trend estimates are likely to be revised when original estimates for subsequent months become available. The approximate effect of possible scenarios on trend estimates are presented below.

Generally, the greater the volatility of the original series, the larger the size of the revisions to trend estimates. Analysis of the building approval original series has shown that they can be revised substantially. As a result, some months can elapse before turning points in the trend series are reliably identified.

The graphs and tables which follow present the effect of two possible scenarios on the previous trend estimates: that the November seasonally adjusted estimate is higher than the October estimate by 4% for the number of private sector houses approved and 13% for other dwelling units approved; and that the November seasonally adjusted estimate is lower than the October estimate by 4% for the number of private sector houses approved and 13% for other dwelling units approved. These percentages were chosen because they represent the average absolute monthly percentage change for these series over the last ten years.

PRIVATE SECTOR HOUSES APPROVED

WHAT IF NEXT MONTH‘S SEASONALLY ADJUSTED ESTIMATE: TREND AS PUBLISHED (1) rises by 4% on Nov 2003 (2) falls by 4% on Nov 2003 no. % change no. % change no. % change 2003 July 10,234 2.2 10,239 2.2 10,264 2.3 August 10,356 1.2 10,359 1.2 10,371 1.0 September 10,407 0.5 10,401 0.4 10,368 - October 10,420 0.1 10,423 0.2 10,314 -0.5 November 10,400 -0.2 10,445 0.2 10,239 -0.7 December - - 10,432 -0.1 10,121 -1.2 - nil or rounded to zero (including null cells)

WHAT IF NEXT MONTH‘S SEASONALLY ADJUSTED ESTIMATE: TREND AS PUBLISHED (1) rises by 13% on Nov 2003 (2) falls by 13% on Nov 2003 no. % change no. % change no. % change 2003 July 4,671 2.1 4,663 2.1 4,703 2.4 August 4,812 3.0 4,809 3.1 4,830 2.7 September 4,947 2.8 4,957 3.1 4,904 1.5 October 5,050 2.1 5,059 2.1 4,884 -0.4 November 5,121 1.4 5,118 1.2 4,787 -2.0 December - - 5,145 0.5 4,646 -2.9 - nil or rounded to zero (including null cells)