Retail sales, Great Britain: September 2016

In September 2016, the quantity bought (volume) of retail sales is estimated to have increased by 4.1% compared with September 2015; all store types except textile, clothing and footwear stores showed growth with the largest contribution coming from non-store retailing.

There was no change in the quantity bought compared with August 2016; decreases in food stores, other stores and textile, clothing and footwear stores were offset by increases in department stores, household goods stores and non-store retailers.

The underlying pattern in the retail sector continues to show relatively strong growth with the 3 month on 3 month movement in the quantity bought increasing by 1.8%.

Average store prices (including petrol stations) fell by 1.1% in September 2016 compared with September 2015; there were falls in average store price across all store types, except textile, clothing and footwear stores and petrol stations. This is the smallest decrease since August 2014.

The amount spent (value) in the retail industry increased by 2.9% compared with September 2015 and increased by 0.1% compared with August 2016.

The amount spent online increased by 22.0% compared with September 2015 and by 2.8% compared with August 2016.

From January 2017 we are improving the way we publish economic statistics, with related data grouped together under new “theme days”. This will increase the coherence of our data releases and involve minor changes to the timing of certain publications. For more information see Changes to publication schedule for economic statistics.

This bulletin presents estimates of the quantity bought (volume) and amount spent (value) in the retail industry for the period 28 August 2016 to 1 October 2016. Unless otherwise stated, the estimates in this release are seasonally adjusted.

The estimates are based on a monthly survey of 5,000 retailers, including all large retailers employing 100 people or more and those with annual turnover of greater than £60 million who employ 10 to 99 people. It is estimated that this survey covers approximately 95% of all known retail turnover in Great Britain.

The quality of the estimate of retail sales

Retail sales estimates are produced from the Monthly Business Survey – Retail Sales Inquiry (RSI). The timeliness of these retail sales estimates, which are published just 3 weeks after the end of each trading period, makes them an important early economic indicator. The industry as a whole is used as an indicator of how the wider economy is performing and the strength of consumer spending. Current price non-seasonally adjusted data are revised for the previous 13 published periods. More information about the data content for this release can be found in the background notes.

Revisions are an inevitable consequence of the trade-off between timeliness and accuracy. The response rate in September 2016 was 62.7% of questionnaires, accounting for 80.0% of registered turnover in the retail industry. Therefore, the estimate is subject to revisions as more data become available.

All estimates, by definition, are subject to statistical uncertainty and for the retail sales index we publish the standard error associated with the non-seasonally adjusted estimates of year-on-year and month-on-month growth in the quantity bought as a measure of accuracy. More information on these standard errors can be found in the background notes and in the quality tables of this release.

We are continually working on methodological changes to improve the accuracy of the retail sales estimates; progress on these can be found on the continuous improvement page.

The datasets offer different ways to access the data, they include:

• non-seasonally adjusted and seasonally adjusted volume and value indexes by industry
• year-on-year and month-on-month growth rates by industry

At a glance

In September 2016:

the quantity bought in the retail industry (volume):

• increased by 4.1% compared with September 2015; the 41st consecutive period of year-on-year growth
• showed no growth compared with August 2016

the amount spent (value):

• increased by 2.9% compared with September 2015
• increased by 0.1% compared with August 2016

In the 5 week reporting period during September 2016, the amount spent in the retail industry was £36.0 billion (non-seasonally adjusted).

This compares with:

• £28.7 billion in the 4 week reporting period for August 2016
• £35.0 billion in the 5 week reporting period for September 2015

This equates to an average weekly spend of:

• £7.2 billion in September 2016 unchanged from August 2016 and
• £7.0 billion in September 2015

Main points

In September 2016:

• all store types except textile, clothing and footwear showed increases in the quantity bought compared with September 2015

• all store types except textile, clothing and footwear, and household goods stores showed increases in the amount spent compared with September 2015

• non-seasonally adjusted data show that the prices of goods sold in the retail industry (as measured by the implied price deflator) decreased by 1.1% compared with September 2015; this was the 27th consecutive month of year-on-year price falls

• compared with August 2016 average store prices have increased by 0.8%, with the largest increase seen in textile, clothing and footwear stores (4.1%)

• compared with September 2015 there were falls in average store prices across all store types, except textile, clothing and footwear stores and petrol stations; however, the fall of 1.1% in all retailing is the smallest fall we have seen since August 2014

More information on how the implied price deflator and other estimates in this release are calculated can be found in section 2 part iii of the background notes.

Table 3 shows the 3 different store types, respective weights in retail sales, year-on-year seasonally adjusted growth rates and average store price.

In September 2016, there were falls of 5.4% in the quantity bought and 5.0% in the amount spent in textile, clothing and footwear stores compared with September 2015. We can see from Table 3 that the fall is due to clothing stores where the quantity bought decreased by 6.4% and amount spent by 5.9%, while there were increases in both textile and footwear stores.

Clothing is by far the largest component, accounting for approximately 88% of the total, while textiles and footwear account for approximately 2% and 10% respectively. Clothing therefore dominates estimates in this sector, thus the deflation in textiles is considerably outweighed by the inflation in clothing stores. Clothing was a contributing factor to the rise in CPI (Consumer Prices Index) where inflation now stands at the highest rate since November 2014.

Figure 1 shows the quantity bought, amount spent and average store price in clothing stores over a longer time period and shows that the monthly time series is fairly volatile. The amount spent in store will be dependent on both the quantity bought and the price, thus changes in one component or both will affect the amount spent. However, through removing the impacts of price and concentrating on the quantity bought, we can still see how the price influences the quantity bought. That is, when prices fall, the quantity bought increases and when price rises the quantity falls.

Looking at the most recent period compared with the previous period, average prices in store increased by 4.3% in September 2016 while the quantity bought decreased by 2.8% and amount spent decreased by 2.1%. However, we saw the opposite in July 2016, when there was an increase in the quantity bought and amount spent as average prices in store fell.

Average store prices are not the only factor that can impact upon sales of clothing; unseasonal weather can be another cause. Feedback from retailers suggests the weakness in September 2016 could be due to the weather during the month impacting on clothing ranges, with consumers delaying purchases of clothing from autumn and winter ranges. The Met Office summary for September 2016 said that the temperature was above the long-term average, making it the equal second warmest September in a series from 1910.

The monthly path for clothing is volatile with erratic month-on-month movements. To remove the volatility the 3 month on 3 month path is shown alongside the monthly series in Figure 2.

In September 2016, there was a fall of 2.8% compared with August 2016, however, looking at the 3 months of July 2016, August 2016 and September 2016, there was a decrease of 0.1% compared with the previous 3 months. This suggests that even though the picture for September 2016 looks bleak for clothing retailers, the 3 month on 3 month movement in the quantity bought shows a flatter picture.

Seasonally adjusted internet sales data are published in the Retail Sales Inquiry (RSI) internet tables and include:

• a seasonally adjusted value index
• year-on-year and month-on-month growth rates

Internet sales are estimates of how much was spent online through retailers across all store types in Great Britain. The reference year is 2013=100.

Main points:

• average weekly spending online in September 2016 was £931.0 million; this was an increase of 22.0% compared with September 2015

• the amount spent online accounted for 15.0% of all retail spending, excluding automotive fuel, compared with 12.6% in September 2015

Table 4 shows the year-on-year growth rates for total internet sales by sector and the proportion of sales made online in each retail sector.

The retail industry is divided into 4 retail sectors:

• predominantly food stores (for example, supermarkets, specialist food stores and sales of alcoholic drinks and tobacco)

• predominantly non-food stores (for example, non-specialised stores such as department stores, textiles, clothing and footwear, household goods and other stores)

• non-store retailing (for example, mail order, catalogues and market stalls)

• stores selling automotive fuel (petrol stations)

Figure 3 shows that for every pound spent in the retail industry:

• 40 pence was spent in food stores

• 43 pence in non-food stores

• 8 pence in non-store retailing

• 9 pence in stores selling automotive fuel

Using these as weights, along with the year-on-year growth rates, we can calculate how each sector contributed to the total year-on-year growth in the quantity bought.

In September 2016 compared with September 2015, all main retail sectors saw an increase in the quantity bought (volume) and amount spent (value). The largest contribution in the quantity bought and amount spent came from non-store retailing.

Table 5 shows how sales varied among different-sized retailers. It shows the distribution of reported change in sales values of businesses (from the RSI sample), ranked by size of business (based on number of employees).

Businesses with 40 to 99 employees saw the largest growth in the amount spent in September 2016 compared with September 2015 (23.6%). Businesses with 100 and over employees showed an increase of 2.5%.

More information on the performance of the retail industry by store type and size can be found in the Business Analysis dataset.

In September 2016, annual growth in retail sales remained at a historically high rate and was unchanged from August 2016, continuing the trend seen in the last few years. Growth continues to be driven by a broad-based increase in all 4 categories (food, non-food, non-store retailing and petrol) but more notably in food sales.

Figure 5 compares a rolling 3 month period with the same period in the previous year and highlights that the volume of retail sales started to grow strongly from mid-2013. The latest data show retail sales growth of 5.4% in the 3 months to September 2016, unchanged from the 3 months to August 2016, the first flat period of growth since the start of the year. The rolling 3 month on 3 month a year ago growth in retail sales has averaged 4.5% since the start of 2016, which is slightly lower than the 2015 calendar year average of 4.6%.

Three distinct periods emerge from Figure 5. Between September 2007 and July 2008, retail sales volumes were experiencing continuous growth, although to a different degree. Growth in inflation (Consumer Prices Index CPI) was lower than average weekly earnings over most of this period; which resulted in rising real earnings, an indicator of the purchasing power of consumers. Moreover, between September 2007 and July 2008, consumer credit increased by 8.6%, which may have been a factor driving retail sales growth.

However, between August 2008 and May 2013, the volume of retail sales fluctuated between periods of contraction and expansion, which may be partly explained by the economic climate over this period and coincided with a reduction in consumer credit of 24.8%. Moreover, growth in average weekly earnings was lower than inflation over most of the period, which implies that earnings fell in real terms.

The most recent data show a notable pickup in underlying sales volumes. Between June 2013 and September 2016, the price level (shown by the implied deflator) fell by 5.2%, coinciding with 14.2% growth in the volume of retail sales over this period. In addition, this upturn in spending has been accompanied by a decline in the savings ratio, from an average of 8.6% over the period 2008 to 2012, to an average of 6.4% from Quarter 1 (Jan to Mar) 2013 to Quarter 2 (Apr to June) 2016. Moreover, since June 2013, consumer credit has followed a broadly upward trend, growing by 20.0% between June 2013 and August 2016.

However, prices have started to rise again – in the year to September, growth in the implied retail sales deflator stood at negative 1.1% – with the higher inflation rate mirroring latest developments in the CPI.

Figure 6 breaks down the growth in total retail sales into the contributions made by food and non-food stores (which include department stores, other stores, clothing, and household goods), non-store retailing (that is, mail orders), and petrol, between September 2007 and September 2016. In the 9 year period, non-store retailing has, on average, made the largest contribution to growth and is the only component to have made a consistently positive contribution to retail sales growth. In the most recent period, all 4 components have made positive contributions to growth, with predominantly food and non-food stores contributing the most and petrol making sizeable contributions since the end of 2014 at a time of historically low oil prices.

In the latest month, growth continues to be driven by a broad-based increase in all 4 categories but more notably in food and non-food stores, contributing 1.8 percentage points and 1.7 points respectively to the 5.5% growth in retail sales in September 2016. Although food growth has been volatile in the last 9 years, there has been a continuous increase of growth over the recent months. This upward trend in food sales is supported by the British Retail Consortium and KPMG suggesting it has been the best quarter for food since 2013.

The latest estimates for retail trade was published by the US Census Bureau on 14 October 2016 in its advanced retail sales estimates for September 2016. They include the amount spent in the US retail industry, including motor vehicles and parts, and food services.

The latest estimates of the volume of retail trade across the European Union, published by Eurostat on 5 October 2016 for August 2016, show the seasonally adjusted volume of retail trade in both the euro area (EA19) and EU28 when compared with July 2016. Note that an accurate comparison cannot be made as Eurostat data are calculated on a 2010 = 100 basis, while data for Great Britain are calculated on a 2013 = 100 basis.

The Retail sales Quality and Methodology Information document contains important information on:

• the strengths and limitations of the data and how it compares with related data
• users and uses of the data
• how the output was created
• the quality of the output including the accuracy of the data

1. Quality

i. Basic quality information

The standard reporting periods can change over time due to the movement of the calendar. Every 5 or 6 years the standard reporting periods are brought back into line by adding an extra week. For example, January is typically a 4 week standard period but January 1986, 1991, 1996, 2002, 2008 and 2014 were all 5 week standard periods. The non-seasonally adjusted estimates will still contain calendar effects. If the non-seasonally adjusted estimates are used for analysis, this can lead to a distortion depending on the timing of the standard reporting period in relation to the calendar, previous reporting periods and how trading activity changes over time.

The non-seasonally adjusted series contain elements relating to the impact of the standard reporting period, moving seasonality and trading day activity. When making comparisons, you should focus on the seasonally adjusted estimates as these have the systematic calendar-related component removed. Due to the volatility of the monthly data, growth rates should be calculated using an average of the latest 3 months of the seasonally adjusted estimates.

When interpreting the data, the relative weighted contributions of the sectors in the “all retailing” series should be considered. Based on SIC 2007 data, total retail sales consists of: predominantly food stores 40.4%, predominantly non-food stores 42.6%, non-store retailing 7.6% and automotive fuel 9.4%.

ii. Standard error

Standard errors determine the spread of possible movements and are a means of assessing the accuracy of the non-seasonally adjusted month-on-month and year-on-year estimates of all retail sales volumes. The lower the standard error, the more confident we can be that the estimate is close to the true value for the retail population.

The standard error of year-on-year movement for “all retailing” in September 2016 is 0.8%. The same level was observed in September 2015. This is the fifth consecutive month that the median of the standard error in the 12-month movement has remained at 0.8%. In the last 12 months, values mostly remained at this level with other fluctuations occurring in August 2015 and December 2015 through to April 2016 when there was a median of 0.9%.

Table 6 shows the year-on-year movement for the non-seasonally adjusted chained volume measure alongside the standard error across the published sector breakdowns for September 2015 and September 2016. The differences between September 2015 and September 2016 highlight that the standard error has only increased for “household goods stores” while it has decreased or remained stable for other sector breakdowns. The greatest decreases are seen for “automotive fuel”, “non-specialised stores”, and “other stores”.

More information on standard errors can be found in the Retail sales quality tables datasets, which are part of this release.

iii. Revisions triangles

Revisions to data provide one indication of the reliability of main indicators. Table 7 shows summary information on the size and direction of the revisions made to the volume data covering a 5 year period. Note that changes in definition and classification mean that the revisions analysis is not conceptually the same over time.

The data section of this bulletin provides these estimates and the calculations behind the averages in the table.

2. Methods

An overview of the Retail Sales Index and a number of methodological articles are also available.

i. Composition of the data

Retail sales estimates are based on financial data collected through the monthly Retail Sales Inquiry. Response rates at the time of publication are included for the current month and the 3 months prior. The response rates for those historical periods are updated to reflect the current level of response, incorporating data from late returns. There are 2 response rates included with a percentage for the amount of turnover returned and another percentage for the amount of questionnaire forms. Historical response rates are available in the quality information dataset.

ii. Seasonal adjustment

Seasonally adjusted estimates are derived by estimating and removing calendar effects (for example, Easter moving between March and May) and seasonal effects (for example, increased spending in January as a result of Christmas) from the non-seasonally adjusted (NSA) estimates. Seasonal adjustment is performed each month and reviewed each year, using the standard, widely used software, X-13-ARIMA-SEATS. Before adjusting for seasonality, prior adjustments are made for calendar effects (where statistically significant), such as returns that do not comply with the standard trading period (there is more information in the Methods, Calendar effects section), bank holidays, Easter and the day of the week on which Christmas falls.

The data collected from the retail sales survey estimate the amount of money taken through the tills of retailers; these are non-seasonally adjusted data. These data consist of 3 components:

• “trend” which describes long-term or underlying movements within the data

• “seasonal” which describes regular variation around the trend, that is, peaks and troughs within the time series (the most obvious is the peak in January and the fall in February)

• “irregular” or “noise”, for example, deeper falls within the non-seasonally adjusted series due to bad weather impacting on retail sales

To ease interpretation of the underlying movements in the data, the seasonal adjustment process estimates and removes the seasonal component. It leaves a seasonally adjusted time series made up of the trend and irregular components.

In the non-seasonally adjusted RSI we see large rises in January each year and a fall in the following February, but these are not evident in the seasonally adjusted index. This peak in January is larger than the subsequent fall, but the trend and irregular components in both months are likely to be similar. This means that the movements in the unadjusted series are almost completely a result of the seasonal pattern.