As advised in Electronic Card Transactions: February 2015, we are revising previously published electronic card transaction data. We are also introducing our new seasonal adjustment tool for the Electronic Card Transaction (ECT) series. These changes will take effect from March 2015 (to be released on 13 April 2015).
Download revised back series
To be ready for the release of electronic card transaction data on 13 April 2015, you can download back data that incorporates these changes up to the February 2015 month from:
Subject category: Economic Indicators
Group: Electronic Card Transactions (ANZSIC06) – ECT
Revised back series for Electronic Card Transactions: February 2015 month (Excel, 6 sheets) from 'Available files' on this page. If you have problems viewing the files, see opening files and PDFs.
Revision to electronic card transaction data
One of our commercial data providers has changed their method for producing their input dataset.
They have provided data to us using this new method back to the May 2010 month.
For data before May 2010, we estimated the value and volume of electronic card transactions by applying a ratio to the previously published figures.
These changes affect all published ECT series, with the largest change in the fuel retailing industry.
The ECT series now provide a more accurate picture of the value and volume of electronic card transactions in New Zealand.
The key changes to the data result from:
- including only purchase transactions – excluding other transaction types like pre-authorisations
- including only approved transactions – excluding declined transactions
- including more accurate e-commerce transactions.
Our new seasonal adjustment tool – X-13ARIMA-SEATS
We will use the X-13ARIMA-SEATS seasonal adjustment tool for ECT from March 2015.
The seasonal adjustment tool we used previously (X-12-ARIMA) is no longer supported by its designers, the US Census Bureau. The change to X-13ARIMA-SEATS ensures continued support, includes improved functionality, and is in line with international best practice.
X-13ARIMA-SEATS uses the same general methodology as X-12-ARIMA. However, some minor improvements are made to the way trading day effects are accounted for, and how the weights (ie measure of importance) are calculated for extreme values. These methodology improvements have contributed small changes to the previously published ECT series.
The change to X-13ARIMA-SEATS affects all seasonally adjusted and trend series for ECT.
New seasonally adjusted quarterly fuel series
We will also introduce a new seasonally adjusted series for quarterly fuel retailing from the release on 13 April 2015.
Historically, quarterly sales for the fuel retailing industry has not shown a reliable seasonal pattern, and has therefore not been seasonally adjusted.
However, from mid-2009 a seasonal pattern emerged in the quarterly value of fuel retailing sales. We will apply seasonal adjustment from the June 2009 quarter onwards for sales values. The earlier portion of this series (from the December 2002 quarter to the March 2009 quarter) will remain unadjusted, as it does not display a seasonal pattern.
Note that this change affects only the quarterly fuel series, and does not affect the seasonally adjusted total retail or total ECT series (which are calculated directly). The monthly fuel series is not affected by this change.
This ‘split’ seasonal adjustment of the quarterly ECT fuel series is consistent with the treatment of fuel sales values in the Retail Trade Survey.
Effect of data revisions
These graphs show the effect of the revisions on actual data.
Figure 1 shows that the value of monthly actual core retailing is about 1 percent lower than the originally published series.
Figure 2 shows that the revised value of monthly actual fuel transactions is up to 17 percent lower than the previously published series.
Figure 3 shows that the revised monthly value of retail transactions is about 3 percent lower than the previously published series.
Figure 4 shows that the revised value of total monthly transactions is about 3 percent lower than the previously published series.
Effect of new seasonal adjustment
These graphs show the combined effect of the revisions and seasonal adjustment changes on seasonally adjusted data.
Figure 5 shows that the revised value of seasonally adjusted monthly transactions for core retailing is about 1 percent lower than the originally published series.
Figure 6 shows that the revised seasonally adjusted monthly value of retail transactions is about 3 percent lower than the previously published series.
Figure 7 shows that the revised seasonally adjusted monthly value of total transactions is about 3 percent lower than the previously published series.
Figure 8 shows that the revised seasonally adjusted quarterly value of fuel transactions is up to 19 percent lower than the previously published series. It also shows that the introduction of a 'split' seasonal adjustment from June 2009 has smoothed out the seasonal fluctuations in the later part of this series. The ‘split period’ line on the quarterly fuel retailing graph marks the point at which the new fuel seasonal adjustment begins.
For more information contact:
Craig Liken or Tehseen Islam
Christchurch 03 964 8700
Published 7 April 2015