The Strange Adjustments Boosting US Data in 2023

Data: US Census Bureau; Diagram: Axios Visuals

You may recall last week’s news of a blockbuster 3% rise in retail sales in January that spilled over global markets. But that report actually showed that retailers made $121 billion less in sales in January than in December — a 16% decline.

• The difference is the result of the seasonal adjustment process applied to most important data – and right now it can send misleading signals about how the economy is shaping up as we enter 2023.

Why it matters: A slew of hot growth and inflation numbers rocked markets this month. But at least some of that heat appears to be due to shifts in seasonal patterns, making this winter’s numbers look more lavish than they are.

• There’s little doubt that the economy has gained momentum so far this year, but it’s less clear how much of it is real.

How it works: If statistical offices did not make seasonal adjustments, the numbers they produce would be highly misleading.

• For example, every November and December there would be an apparent boom in consumer spending and attitudes tied to the holiday season — and then a depression every January as sales fall back to earth and seasonal aid is let go.
• However, these adjustments are based on previous results, meaning they’re slow to catch on as seasonal patterns change.

So for example If the 2020-21 pandemic supply shortages caused people to start their Christmas shopping earlier than usual, the seasonal adjustment would exaggerate the strength of October’s retail sales numbers, depressing November and December.

• It would also make the January numbers look much stronger than reality as the fall in spending from December to January would be less pronounced than the seasonal models predict.
• This is exactly what appears to have happened in last week’s retail sales, which were seasonally adjusted negative in November and December and then significantly positive in January.

• Total retail sales in January were up 0.7% on October and seasonally adjusted more consistent with steady gains than a January boom.

Between the lines: Weather can amplify seasonal distortions. On a normal January, freezing temperatures and snowstorms disrupt economic activity across much of the country. Seasonal adjustments take this into account.

• But this was an unusually warm winter, meaning seasonal adjustments are increasing reported activity above the true underlying trend.
• A San Francisco Fed model that adjusts reported job numbers for weather effects found that the nation would have added about 390,000 jobs in January — not the 517,000 the Labor Department reported — had it not been for a warm winter.

The bottom line: “Right now, it’s difficult to determine whether COVID-induced changes in consumer behavior and business practices are altering seasonal data adjustments, or whether real underlying economic activity is as strong as some recent economic indicators suggest,” said Doug Duncan, chief economist at Fannie Mae.