Data for the Retail Sales Index (RSI) is collected monthly from a sample of 1,600 retail enterprises. Despite the current difficult circumstances, many retailers have continued to supply us with this information and we wish to thank all survey respondents for their support.
However, due to temporary closure and the difficulties faced by businesses, the response rate for the March 2020 RSI was about five percentage points lower than would normally be the case. The March response covered about 53% of total retail sales in the economy.
The Retail Sales index is based on a matched sample of enterprises in consecutive months. Imputation is not normally carried out for smaller enterprises which have not returned data. In each sector key companies are identified based on their significance to the sector (turnover). If key company data is not received on time for the Provisional results, then an estimate is made using trends from other similar companies and other data available at the time.
The reduced response rate for March 2020 meant that a limited imputation of smaller enterprises was also carried out. This occurred where the response for a sector or a cell (based on turnover size) was low. Estimates for these missing smaller enterprises were made using the trends shown by responding enterprises in the same sector and, in some cases, bespoke imputation using auxiliary data sources was used.
The auxiliary data sources have also been used to validate the results for March 2020 at retail business category level. The CSO is satisfied that this full range of information supports the trends estimated in the Retail Sales Index for March. The auxiliary sources used for imputation and validation included:
The March figures, and subsequent months, are an exception to existing seasonal patterns. Seasonally adjusting the Retail Sales Index will be challenging until the scale and shape of the impact COVID-19 has on the time series is better understood. For the time being March 2020 is modelled as an additive outlier which will be reviewed as required. Users should be aware that further observations, as data becomes available in the months ahead, and revisions to the seasonal adjustment models may result in revisions to the seasonally adjusted series.
Most establishments were open for either a part of or most of March. Data collection for April will prove to be more challenging as the range of outlets open for business has diminished to only those deemed to be essential services, such as supermarkets, other food outlets, fuel and pharmacies.
The CSO wishes to thank all those enterprises who supplied their retail sales data for March and urge them to continue to do so for April, even where turnover for many enterprises will be nil. Such information is still needed to measure the precise impact of the crisis on the retail sector. The CSO will also continue to monitor the range of supplementary information described above to derive the best estimate for retail sales in April. The April Retail Sales Index is planned for publication on 29 May.
Stephanie Kelleher (+353) 21 453 5123