Monthly retail sales is one of the most closely watched economic releases and one of the most consistently over-interpreted. It is volatile, heavily revised, measured in nominal terms, and covers a narrower slice of consumer spending than its name implies. Each of those is a reason to treat a single month with caution.
It is a nominal series
The headline figure measures dollars spent, not goods bought. It is not adjusted for inflation.
This single fact accounts for a large share of the misreading. In a period of meaningful goods inflation, retail sales can rise every month while the actual volume of goods purchased falls. Consumers are spending more and receiving less. The headline describes that as growth.
The correction is to compare the series against goods price inflation over the same period. If nominal sales are growing at three percent and goods prices are up four, real consumption is contracting. Coverage almost never performs that subtraction.
It excludes most services
Retail sales captures goods and food services. It largely excludes the rest of the service economy — healthcare, education, housing, insurance, travel other than eating out, professional services.
Services are the majority of consumer spending in developed economies, and the majority by a wide margin. So the release covers a minority of consumption and is routinely reported as a read on “the consumer.”
This matters most when spending is rotating between goods and services. A shift from buying things to buying experiences shows up as weak retail sales alongside perfectly healthy total consumption. The headline records a slowdown that is really a composition change.
The volatile components
Three categories generate a disproportionate share of month-to-month movement.
- Motor vehicles. Large-ticket, lumpy, and sensitive to incentives and supply. A single strong or weak month for auto sales can swing the headline substantially without indicating anything about broader spending.
- Petrol stations. Because the series is nominal, a change in fuel prices moves this component directly. A rise in petrol prices increases measured retail sales while reducing the money households have available for everything else.
- Building materials. Driven by construction and weather rather than consumer sentiment.
This is why analysts focus on the “control group” measure, which strips out autos, fuel, building materials and food services. It is a smaller series, but it is the one that feeds the consumption component of national accounts, and it is considerably less noisy. When the headline and the control group diverge, the control group is usually the better signal.
Seasonal adjustment does heavy lifting
Raw retail sales swing enormously across the year. December is vastly larger than January in unadjusted terms. Everything published is seasonally adjusted, and the adjustment factors are large.
When the adjustment is large, small errors in it produce large errors in the published figure. Two situations reliably cause trouble:
- Shifting holiday timing. Holidays that fall on different dates each year, and the number of weekends in a reporting month, both distort the comparison in ways the adjustment handles imperfectly.
- Weather. Seasonal adjustment accounts for a typical February, not an unusually severe one. Weather-driven weakness is real spending that was deferred, and it typically reverses the following month — producing a weak print followed by a strong one, neither of which means much.
The revisions are large
Like most monthly indicators, the first estimate is built on incomplete survey responses and is revised as more arrive. Revisions to retail sales are frequently large enough to change the direction of the reported month.
A month initially reported as a decline can be revised to a gain. By the time that happens the original figure has already been traded on, written about and folded into forecasts, and the revision receives a fraction of the attention. This is a general property of monthly economic data rather than a quirk of this series, and it is worth internalising as a habit — the same discipline applies to the employment report, as we set out in why the jobs number gets revised twice.
Reading it properly
- Use the control group, not the headline.
- Deflate it. Compare nominal growth against goods price inflation to get at volume.
- Use a three-month average. The month-to-month series is dominated by noise; the trend is where the signal is.
- Check the revision to the prior month before reacting to the current one.
- Remember what is missing. For a view on total consumption, the personal consumption expenditures data is the more complete series, though it arrives later.
None of this makes the release useless. It is timely, it is one of the earliest reads on demand in any given month, and over a quarter or two it tracks consumption reasonably well. The error is treating a single month as information about the direction of the economy, when the data was never precise enough to support that.
