Retail Sales YoY

Retail sales YoY is the % change in Japan retail sales value compared with the same month a year earlier. It covers stores selling durable and nondurable goods, counting sales at stores, by mail order and online, with the total including consumption tax. It does not measure services, sales volumes adjusted for prices, or household spending tracked by a separate survey. It is a fundamental survey under the Statistics Act, with rules dating to 1953.

The survey reference date is the end of each month, with stocks, or unsold goods, counted at the end of March, June, September and December. The preliminary report, or early estimate, for a survey month appears around the end of the next month at 08:50 Japan Standard Time. Each release shows wholesale plus retail totals, retail sales by industry and format, original and adjusted measures, and a short assessment sentence. Figures marked p are preliminary and r are revised later, and readers weigh YoY direction with adjusted month on month change from the prior month.

Sales move with consumer demand and tax changes across 7 industry groups merged into 6 points.

• General Merchandise: broad everyday goods sold in general stores.
• Fabrics Apparel and Accessories: clothing, fabrics and personal accessories stores.
• Food and Beverages: food and drink retail.
• Motor Vehicles: automobile retail, surveyed individually regardless of store size.
• Machinery and Equipment: home appliances, office machinery and similar equipment stores.
• Fuel: fuel retail, surveyed individually regardless of store size.

A teaching example is the consumption tax rise on 1 April 2014, when the rate rose from 5% to 8%. March 2014 sales soared 11.0% from a year earlier, released 27 April 2014, led by machinery and equipment at 37.5% and general merchandise at 19.6%. Autos reached 9.4%, fuel 5.3% and food and beverages 5.1%, then April 2014 fell 4.3% with a 13.7% slide from the prior month. The mirror-image swing teaches that a lone YoY print can misstate trend strength or weakness, and seasonal adjustment flags such months as explicit outliers.

1. Sampling cell ratio: Each sampling cell, or grouping by industry and size, applies its sample change ratio to the prior published total. For example, sample sales of 900 then 990 give a ratio of 1.1, so a prior total of 5000 becomes an estimate of 5500.
2. Enterprise census formats: Department stores and supermarkets add collected totals directly instead of estimates. Convenience stores, large electronics sellers, drugstores and home improvement chains do the same. Each of these formats is surveyed as a census, or full count, not a sample, and past levels link to censuses.
3. Year-on-year change: The year-on-year % compares a month with the same month of the prior year. It takes 100 times the difference over the prior year value. For example, March sales of 5500 against 5000 a year earlier give 10.0%.
4. Seasonal adjustment and index base: Seasonal adjustment, or removal of calendar effects, uses X-12-ARIMA, a United States Census Bureau method. It divides the original index by seasonal and trading day factors. Stated models include (210)(011) for retail and for general merchandise. Original indices set the 2015 calendar year average at 100, a base used since January 2017.
5. Sampling design: The design holds sampling error to 5% or less for each retail category and 8% or less for wholesale. Smaller stores with fewer than 20 workers come from 143 survey areas. The large formats above face a full census instead.