During the March 2011 quarter, the Christchurch earthquake on 22 February 2011 caused significant damage to the city of Christchurch, particularly the central and eastern parts of the city. Work has been done with the Retail Trade Survey to assess the impact of the earthquake on retail trade in Christchurch city and the Canterbury region. Despite the earthquake, the Retail Trade Survey data has remained robust. In this note we explain the impact of this on the survey and what we have done to account for this in the survey.
The Retail Trade Survey is a sample survey of retail businesses. It is designed to produce an estimate of consumption expenditure each quarter. Because the Retail Trade Survey is a sample survey, responses by businesses are weighted to represent those of similar businesses. The Retail Trade Survey is designed to produce accurate estimates of spending at the national and industry level (for 15 industry groups). Estimates are available at regional levels, but the sample error associated with these is higher. The Retail Trade Survey is not designed to produce accurate estimates at the regional level, although there is a reasonably good sample of businesses included in significant urban regions, including Canterbury.
It was expected that there may have been issues with the following in the Retail Trade Survey for the March 2011 quarter:
- response rates – how many businesses responded to the survey
- imputation rates – how many businesses’ results needed to be estimated because they did not respond to the survey
- sales values – the value of sales estimated in the quarter
- stock estimates – the value of stocks estimated at the end of the quarter.
Response rates: Although it was anticipated that the earthquake would have potentially affected the response rates to the quarterly Retail Trade Survey, there was still a very good response rate achieved for Canterbury and Christchurch, with over 85 percent of surveyed respondents able to provide information. Although this is slightly lower than normal, it is still statistically sound.
Imputation rates: In order to make an accurate estimate of the sales and stock levels, there needs to be a sufficiently high number of responses in each industry to help impute (estimate) information for those businesses that are unable to respond to the questionnaire. Although it was anticipated that it would be a challenge to get responses in Christchurch during this period, the imputation rate was not significantly higher than normal.
Sales values: It was expected that there may be significantly lower sales values than normal for a number of businesses which may have affected the sales levels in the survey across New Zealand. Sales levels were definitely lower for a number of the businesses in the Retail Trade Survey for the March 2011 quarter. A number of businesses were not able to trade for almost half of the survey period (being open for 53/90 days of the quarter).
Despite this, the number of businesses that were able to provide information was very high, and the level of sales recorded in this period was not significantly lower than other periods. Although there were lower sales for a number of businesses in central and eastern Christchurch, there were also a number of businesses in other parts of the city and region that recorded offsetting increased sales. The areas of spending where this was evident included consumable goods (groceries) and fuel.
Of the respondents in the Retail Trade Survey, the proportion from Canterbury that were unable to provide information on the level of sales made was not significantly higher than the numbers unable to respond in a normal quarter.
In analysing the imputed values of the Christchurch non-respondents, there was a possibility that an adjustment would need to be made to reflect the impact of the earthquake on trading in Christchurch or Canterbury. A way of doing this would have been to adjust for the sales levels for those who did not respond to reflect the shorter trading period (53/90 days). This was investigated; however, the actual impact recorded, along with the potential offsetting higher sales recorded in less-damaged parts of the city and region, meant that it was not statistically sound to make this adjustment. After all the earthquakes since September 2010, Christchurch retailers have shown remarkable resilience, and in many cases, have relocated and started trading again very quickly.
Stock levels: An estimate of stock levels is made at the end of each quarter. For a number of Christchurch businesses, it was expected that it would be very difficult to estimate the stock levels, due to not being able to get to them or measure them (particularly if the business was located in the central city, or had been destroyed). Although this was the case for a number or retailers, there were also a number of retailers who were able to provide stock estimates.
Again, there was not a statistically significant difference between the numbers who were not able to respond this period and those who normally would. In cases where it was clear that businesses were not able to re-open and stocks were not able to be recovered, a stock level of zero was recorded reflecting loss of stocks in this period.
In the national accounts the key stock number is the change in stocks over the period, excluding losses such as stock destroyed by the Canterbury earthquakes. Therefore, stock levels in this release will not be the same as those used to produce the quarterly gross domestic product release, which will be published on 7 July.
Statistics New Zealand would like to thank all respondents for their quarterly retail trade information, especially the many from Canterbury who were able to provide information despite the impact of the 22 February earthquake on their businesses.
Further information, including methodological information on the analysis done, is available on request.
For enquiries contact:
Christchurch 03 964 8700