This document presents the methodology used to produce volume measures of retail trade sales. To achieve this goal, information from the Consumer Price Index (CPI) the Retail Commodity Survey (RCS) and the Monthly Retail Trade Survey (MRTS) are combined to produce monthly estimates for the total retail trade industry.
Changes in current dollar retail sales can be decomposed into two elements: a price element, or the part of the growth linked to price variations and a volume element, which covers the change in quantities and quality of the goods and services sold. The chained dollar and constant price data for retail sales provide two evaluations of the changes in the volume of sales. The volume measures are obtained by removing from the current dollar value of sales the price variations measured by appropriate price indexes. This process is known as deflation.
The price indexes used for deflation come from the Consumer Price Index (CPI – survey no. 2301) program. Adjustments are made to the CPI indexes to exclude changes in retail sales tax since the retail sales data exclude HST, GST or PST1, whereas the CPI measures includes the effect of these tax changes. An unpublished dataset from Consumer Prices Division is used to adjust the CPI to the retail sales concept.
The Retail Commodity Survey (RCS – survey no. 2008) collects information for 120 exhaustive categories of goods sold by type of retail outlet. This survey provides a breakdown of the retail outlet’s total sales by commodity. Each RCS commodity is matched with the most suitable CPI component, or a weighted combination of CPI components. The RCS is the cornerstone of the methodology since its two dimensions allow the transformation of commodity prices into industry prices weighted by commodity sold.
The Monthly Retail Trade Survey (MRTS – survey no. 2406) produces data on retail trade sales in current dollars by type of store. In order to bring consistency to the data, the published estimates from RCS by type of retail outlet are benchmarked at the micro data level to the MRTS results. However, there is one single exception. Department stores (NAICS 452110) differ between the two surveys because RCS includes concession sales while MRTS does not.
Volume estimates of retail sales are published about fifty days after the end of the reference period, but the RCS data that are essential to produce them is available quarterly within 90 days of the reference quarter. Therefore, a projection method is used to derive a commodity breakdown of MRTS data for the most recent months. For months when a RCS breakdown is not available, shares from the most recent month of RCS data serve as a starting point. The previous year’s difference in shares between the month being projected and the month being used is applied to the most recent month available2. The objective of this projection method is to obtain a more up-to-date weighting structure that takes into consideration the seasonality in the goods sold. The calculated shares are applied to the current dollar value of sales by store type provided by the MRTS to derive current dollar value of sales by store type and commodity.
To calculate 2007 dollar volume data, the price indexes are adjusted in a way that the average index equals 100 for this reference year. Current dollar sales by store type and commodity are then divided by their respective price indexes to derive constant price sales. Finally, the sales in volume at constant prices are summed over all commodities to derive volume of sales at constant prices by store type. The volume of total sales at constant prices is the sum of the volumes of sales at constant prices by store type.
The total volume of sales in chained dollars corresponds to the geometric mean of two evaluations of the variations in volume between two consecutive months. The first evaluation is based on the aggregated price of the previous month and the other is based on the price of the current month. Chained dollar estimates for total retail sales are derived from constant dollar estimates of sales by store type using a Fisher index formula. Only the total volume of sales of the retail sector as a whole can be computed in chained dollars.
Implicit price indexes are derived by dividing the current dollar sales by the volume in chained dollars or the volume in constant prices. They can be thought of as the change in the average price of goods sold at retail stores. They reflect both changes in prices and changes in the composition of goods and services sold.
Current dollar sales and implicit price indexes by store type are both seasonally adjusted using the X-12-ARIMA method. Seasonally adjusted volume sales by type of store are derived indirectly by dividing seasonally adjusted current dollar sales by their corresponding seasonally adjusted implicit price indexes by type of store. Seasonally adjusted volumes of sales by type of store are then aggregated to derive seasonally adjusted volumes of total sales.
For example, when data are released for June 2012 the most recent RCS data available was for March of 2012 so the shares were computed as Shares June 2012 = Shares March 2012 + Shares June 2011 – Shares March 2011.