Series
Christchurch Retail Trade Indicator
en-NZStatistics New Zealand
en-NZThe Christchurch retail trade indicator is an experimental series. We release it to provide information on the state of the Christchurch retail trade industry after the earthquake sequence that began in September 2010. The statistics are provisional, because they reflect new methods that we may modify.
en-NZThe purpose of the series is to provide information on the state of the Christchurch retail trade industry following the Canterbury earthquakes that began in September 2010. We are releasing the results as an experimental series while work into the methods is ongoing.
We will publish the series indefinitely, depending on the level of customer demand and the characteristics of the Christchurch recovery.
en-NZ3 Quarterly
Standardising the RTS reference period to quarterly
From October 2003 to December 2010, we released the RTS data monthly. From March 2011, the RTS changed to a quarterly release. To produce data on a consistent basis, we changed the RTS data from October 2003 to December 2010 to a quarterly frequency by aggregating the RTS monthly data for each business to produce quarterly values.
We advise our customers to consider the following when analysing the Christchurch retail trade indicator series.
- We recommend focusing on changes and movements in the series rather than the actual level of activity – the series does not fully cover Christchurch businesses.
- The series is constructed using goods and services tax (GST) sales data from Inland Revenue, supplemented with RTS data. The GST data includes sales of retail goods and services and other income such as sales of capital items and businesses, and insurance payouts resulting from earthquakes (see insurance and depreciation). We are investigating ways to exclude sales of capital items, businesses, and insurance payouts from the GST data. Until this work is complete, we will exclude only large values that can be identified at aggregate levels, and the series will be released on a provisional basis.
- We publish data for two Australian and New Zealand Standard Industrial Classification 2006 (ANZSIC) divisions: G retail trade, and H accommodation and food services. Although we refer to industries below this level in the commentary, the descriptions are only indicative. Data below this published level is not of high enough quality for statistical release.
Coverage
Christchurch Retail Trade Indicator
This data collection covers the broad methodological information for the Christchurch Retail Trade Indicator series.
en-NZMethodology
Period-specific information
Measurement errors
All statistical estimates may have measurement errors that our customers should consider when analysing our statistical outputs. Errors from the Christchurch retail trade indicator series are of two types – model errors and other measurement errors.
Model errors
We use models to standardise the GST reference period to quarterly, which may include model errors. These errors measure the variability that occurs when we apply a statistical model to produce estimates, which quantifies the cumulative effect of model 'imperfections'. Relative model error is a measure of model error that is expressed as a percentage of the estimate at a confidence interval limit. It gives the levels of accuracy of the published estimates.
The table below shows the relative model errors for the level estimates of the Christchurch retail trade indicator series (at the 95 percent confidence interval limit). They show a 95 percent chance that the true value of total retail and hospitality sales in Christchurch for the March 2016 quarter (disregarding the business undercoverage) lies within 0.7 percent of the published estimate.
Model errors for estimates of Christchurch retail trade indicator industries | ||||
---|---|---|---|---|
Quarter | G Retail trade | H Accommodation and food services | Total | |
Percent | ||||
Sep 10 | 0.9 | 1.4 | 0.8 | |
Dec 10 | 0.8 | 1.5 | 0.7 | |
Mar 11 | 1.4 | 3.3 | 1.3 | |
Jun 11 | 0.4 | 1.7 | 0.4 | |
Sep 11 | 0.5 | 2.7 | 0.6 | |
Dec 11 | 0.8 | 1.8 | 0.7 | |
Mar 12 | 0.9 | 2.2 | 0.8 | |
Jun 12 | 0.6 | 1.8 | 0.6 | |
Sep 12 | 0.7 | 2.3 | 0.7 | |
Dec 12 | 0.7 | 2.0 | 0.6 | |
Mar 13 | 0.6 | 1.6 | 0.5 | |
Jun 13 | 0.6 | 1.6 | 0.5 | |
Sep 13 | 0.4 | 1.1 | 0.4 | |
Dec 13 | 0.6 | 1.8 | 0.6 | |
Mar 14 | 0.7 | 2.2 | 0.7 | |
Jun 14 | 0.4 | 1.4 | 0.4 | |
Sep 14 | 0.9 | 1.6 | 0.8 | |
Dec 14 | 0.5 | 2.2 | 0.5 | |
Mar 15 | 0.8 | 1.7 | 0.7 | |
Jun 15 | 0.7 | 1.9 | 0.6 | |
Sep 15 | 0.8 | 2.2 | 0.8 | |
Dec 15 | 0.6 | 1.3 | 0.6 | |
Mar 16 | 0.8 | 1.5 | 0.7 | |
Jun 16 | 0.4 | 1.7 | 0.4 |
Other measurement errors
Other measurement errors arise from inaccuracies in reporting by respondents, errors in recording and coding data, and Retail Trade Survey (RTS) imputation processes. The size of these errors is difficult to quantify. We may revise statistics if we find significant errors in subsequent quarters.
General information
About the Christchurch retail trade indicator
The purpose of the series is to provide information on the state of the Christchurch retail trade industry following the Canterbury earthquakes that began in September 2010. We are releasing the results as an experimental series while work into the methods is ongoing.
We will publish the series indefinitely, depending on the level of customer demand and the characteristics of the Christchurch recovery.
Caution about using data
We advise our customers to consider the following when analysing the Christchurch retail trade indicator series.
- We recommend focusing on changes and movements in the series rather than the actual level of activity – the series does not fully cover Christchurch businesses.
- The series is constructed using goods and services tax (GST) sales data from Inland Revenue, supplemented with RTS data. The GST data includes sales of retail goods and services and other income such as sales of capital items and businesses, and insurance payouts resulting from earthquakes (see insurance and depreciation). We are investigating ways to exclude sales of capital items, businesses, and insurance payouts from the GST data. Until this work is complete, we will exclude only large values that can be identified at aggregate levels, and the series will be released on a provisional basis.
- We publish data for two Australian and New Zealand Standard Industrial Classification 2006 (ANZSIC) divisions: G retail trade, and H accommodation and food services. Although we refer to industries below this level in the commentary, the descriptions are only indicative. Data below this published level is not of high enough quality for statistical release.
Feedback
We welcome feedback. Please email comments to: info@stats.govt.nz
Target population
The target population is all geographic units (GEOs) on Statistics NZ's Longitudinal Business Frame (LBF) that are operating in Christchurch and are classified in ANZSIC06 to:
- retail trade (ANZSIC division G)
- accommodation and food services (ANZSIC division H).
Christchurch GEOs are those classified to the Christchurch city territorial authority. The data uses 2011 territorial authority boundaries. A GEO must belong to an economically significant enterprise (see definitions).
Many Christchurch GEOs were damaged by the 2010 and 2011 earthquakes. We have several sources to identify temporary and permanent business closures. However, most of these sources have a timing lag, which means there is often a delay between a business closing and it 'ceasing' on the LBF.
We chose the LBF as the population source for the series because the Business Frame (which is normally used to select population and samples for Statistics NZ's sub-annual financial indicators) provides only the latest-available data on businesses. However, the LBF records businesses' attributes over time. To provide information on the state of the Christchurch retail trade industry after the Canterbury earthquakes, we produced a suitable back series to allow the seasonal adjustment of data.
Data sources
Data is sourced from the RTS and from Inland Revenue’s GST sales information.
The RTS produces statistics at the national level, by industry. Statistics at a more-detailed level are often subject to higher sampling error. While most large businesses are in the sample, only portions of small to medium-sized businesses are sampled. We therefore cannot expect the sample to completely represent retail businesses in every city and region. If we attempted to produce statistics for Christchurch using only the RTS, estimates would be uncertain. For example, the March 2011 RTS estimated a 12.7 percent sampling error for the movement in Christchurch sales.
To produce more robust estimates, we use GST data from Inland Revenue and supplement it with RTS data – to produce the Christchurch retail trade indicator. This is so we have a greater coverage of Christchurch businesses.
A key difference between the GST and RTS data is the statistical unit from which the data is collected. GST data is collected at the enterprise (or legal entity) level, while RTS data is collected at the GEO level.
We use GST data wherever it is deemed to be suitable for use – for enterprises with one geographic location and involved in one industrial activity. For these enterprises, we need no additional statistical processes to apportion the GST data from the enterprise to the GEO level.
We use RTS data for the remaining businesses – generally large enterprises, so most are included in the RTS sample. These enterprises have one or more of the following characteristics:
- are located at more than one geographic location
- are involved in more than one industrial activity
- are part of a GST group (in which a single business reports on behalf of a number of businesses linked through ownership).
The GST sales variable from which data is sourced has a different definition from the RTS sales variable (see table below). Both the RTS and GST sales data are calculated as 'exclusive of GST' for our use in the series.
Definitions of sales variables in Christchurch retail trade indicator data sources | |
---|---|
Data source | Sales definition |
RTS questionnaire | Respondents provide:
|
GST form | Total sales and income for the period (including zero-rated supplies) |
Undercoverage of Christchurch businesses
The Christchurch retail trade indicator series has an undercoverage of Christchurch businesses. We make no attempt to impute for this undercoverage.
Reasons for undercoverage in Christchurch retail trade indicator | |
---|---|
Source of undercoverage | Description |
Businesses with no suitable GST or RTS data | Businesses with one or more of the following characteristics (so using GST data is not suitable):
|
Late GST filers | Businesses located at one geographic location and involved in one industrial activity whose GST return is not available in time for publication. |
Six-monthly GST filers | We exclude six-monthly GST filers from the series, as we have no method to standardise this filing period to a quarterly frequency. Their contribution is only small. |
Economically insignificant businesses | Businesses that do not meet any of the economic significance criteria (these are generally excluded from our economic surveys). |
We estimated the levels of undercoverage for each series. As the level of undercoverage remains consistent over time, the series can be used to analyse changes and movements in Christchurch retail trade activity.
ANZSIC06 division | Estimated average undercoverage (percent) |
---|---|
G Retail trade | 9 |
H Accommodation and food services | 17 |
Total | 10 |
Standardising the reference periods to quarterly
This section describes the methodology used to standardise the GST and RTS reference periods to a quarterly frequency. We developed the GST methods recently and they may be refined.
Standardising the GST reference period to quarterly
Inland Revenue collects GST data as part of administering New Zealand’s taxation system. It is not primarily designed to produce economic statistics. We needed to develop methods to transform the GST data, which is submitted at different frequencies, to a quarterly frequency.
Enterprises submit GST returns monthly, two-monthly, or six-monthly (depending on the annual turnover of the business). The two-monthly and six-monthly returns can also be filed using different balance months. For instance, the two-monthly 'TA' return is filed using January, March, May, July, September, and November balance dates. The 'TB' return is filed on alternate months (February, April, June, August, October, and December).
Treatment of each of these GST-filing frequencies is described in the table below, along with an example using a typical June quarter.
Treatment for standardising GST reference periods to a quarter | ||
---|---|---|
GST-filing frequency | Treatment | June quarter example |
Monthly | Sum the three months of the quarter | Activity = April + May + June months |
TA | Use the TA return in the quarter that includes two of the three months. We apply a modelling factor based on the activity of monthly returns for the same period. | Activity = May TA return x modelling factor for June |
TB | Use the TB return in the quarter that includes two of the three months. We apply a modelling factor based on the activity of monthly returns for the same period. | Activity = June TB return x modelling factor for April |
Six-monthly | Exclude from Christchurch retail trade indicator series – no method developed to standardise this filing period to a quarterly frequency. |
Standardising the RTS reference period to quarterly
From October 2003 to December 2010, we released the RTS data monthly. From March 2011, the RTS changed to a quarterly release. To produce data on a consistent basis, we changed the RTS data from October 2003 to December 2010 to a quarterly frequency by aggregating the RTS monthly data for each business to produce quarterly values.
Interpreting the time-series data
The following discussion will help data users understand the time series.
Quarters available
Christchurch retail trade indicator series data is available on a quarterly basis back to the December 2003 quarter. To provide information on the state of the Christchurch retail trade industry following the Canterbury earthquakes, we produced a suitable back series before seasonally adjusting the data.
Industry descriptions
Data is published for the following two industries.
Christchurch retail trade indicator series | |
---|---|
ANZSIC06 industries, subdivision codes, and descriptions | |
Industry | Subdivision and description |
G Retail trade |
|
H Accommodation and food services |
|
Publication timeframes
The Christchurch retail trade indicator series is published around three months after the reference quarter. This allows Inland Revenue at least two months of GST-processing time after the reference quarter. We estimate this should allow at least 95 percent of the GST sales value to be included when we construct the Christchurch series.
Delays in receiving GST returns due to late filing by the employer or Inland Revenue processing are possible. Inland Revenue's processing of GST returns can also be affected by updates to systems or administrative changes. We review the quality of the GST data each quarter. Our release will be delayed if necessary to ensure data is of sufficient quality.
In the March 2011 quarter, the GST sales value that we could use in the series, after two months of Inland Revenue processing time, was less than in previous quarters. Presumably this was due to a delay in businesses filing GST returns after the 22 February 2011 earthquake. For this quarter, an extra two months of Inland Revenue processing time was incorporated into the series construction, to ensure the processing rate was comparable with previous quarters.
Revisions policy for actual series
Revisions from the RTS are incorporated into the Christchurch retail trade indicator actual series. (Actual series have no seasonal fluctuations or short-term irregular movements removed.)
Other factors may result in revisions to the Christchurch retail trade indicator actual series.
- The LBF is refreshed on a monthly basis. Updates may result in the birth, death, and reclassification of businesses in the Christchurch retail trade population.
- GST data can be updated by Inland Revenue.
We are assessing the effect of these other factors on the series. Until this work is completed, we will not incorporate other revisions into the Christchurch retail trade indicator actual series.
Movement in the June 2010 quarter
The movement in the June 2010 quarter needs to be treated with caution. This was the first quarter the RTS dataset was available with an ANZSIC06 design, so the level of undercoverage (businesses with no suitable GST or RTS data) decreased. This caused a small upward level shift that affected the June 2010 quarter movement. Although we did not release the official RTS ANZSIC06 series until October 2010, we ran a concurrent ANZSIC96 and ANZSIC06 design from April 2010 to September 2010.
Seasonally adjusted series
We produce the seasonally adjusted and trend series using the X-13-ARIMA-SEATS package developed by the U.S. Census Bureau, to comply with international best practice.
Seasonal adjustment aims to eliminate the impact of regular seasonal events (such as annual cycles in agricultural production, winter, or annual holidays) on time series. This makes the data for adjacent quarters more comparable.
We recalculate all seasonally adjusted figures each quarter. This enables the seasonal component to be better estimated and removed from the series.
Estimated trend
For any series, the survey estimates can be broken down into three components: trend, seasonal, and irregular. While seasonally adjusted series have the seasonal component removed, trend series have both the seasonal and the irregular components removed. Trend estimates reveal the underlying direction of movement in a series, and are likely to indicate turning points more accurately than seasonally adjusted estimates.
We calculate the trend series using the X-13-ARIMA-SEATS seasonal adjustment package. They are based on a five-term or seven-term moving average of the quarterly seasonally adjusted series, with an adjustment for outlying values.
Trend estimates towards the end of the series incorporate new data as they become available and can therefore change as more observations are added to the series. Revisions can be particularly large if we treat an observation as an outlier in one quarter, but it becomes part of the underlying trend as further observations are added to the series. Typically, only the estimates for the most-recent quarter will change substantially.
Comparison with the Retail Trade Survey
The following table provides a summary of the main differences between the Christchurch retail trade indicator series and the RTS. We advise caution when comparing the two.
Main differences between Christchurch retail trade indicator and Retail Trade Survey | ||
---|---|---|
Christchurch retail trade indicator | Retail trade survey | |
Population | Christchurch GEOs classified to ANZSIC06 G and ANZSIC06 H | National GEOs classified to ANZSIC06 G and ANZSIC06 H |
Population source | Longitudinal Business Frame | Business Frame |
Data sources | Retail Trade Survey and GST data from Inland Revenue | Retail Trade Survey(1) |
Frequency | Quarterly | Quarterly |
Published industries | 2 published industries | 15 published industries |
Design | Includes all GEOs, despite business undercoverage (described below) | Sample of GEOs |
Undercoverage |
| Economically insignificant businesses |
Publication timeframe | 3 months after reference quarter | 6 weeks after reference quarter |
Area boundaries | 2011 area boundaries – Christchurch city territorial authority includes Banks Peninsula | 2003 area boundaries – Christchurch city territorial authority excludes Banks Peninsula |
Application | Analysis of changes and movements | Analysis of levels, changes, and movements |
1. RTS does use forecasted estimates for small businesses from GST data. |
en-NZ
Coverage
Christchurch city territorial authority. The data uses 2011 territorial authority boundaries.
en-NZ