We’ve got a data-quality problem

Stay informed with free updates

Simply sign up to the US economy myFT Digest — delivered directly to your inbox.

Coming up with a coherent narrative about the US economy is a fraught exercise today. It’s doing well and everyone hates it. Job markets are cooling but wages are rising. Fast food and rent is too pricey but gas is cheap, and so on.

In other words, the Economic Narrative — and even the outlook for policy — seems to be changing quickly from one economic report to the next. And some of these reports have become less reliable since the pandemic, as Goldman Sachs writes in a note recently made public here.

Retail-sales figures are an especially timely example. Goldman Sachs argues that the Commerce Department’s seasonal-data adjustments haven’t kept up with longer-term changes in consumers’ habits since Covid-19.

To explain in more detail: Everyone knows Americans spent more on stuff and less on services during the pandemic, building remote-work set-ups and doing home-improvement projects. What’s more surprising is that the shift hasn’t fully reversed itself, even since vaccines and back-to-office mandates were introduced. This handy chart from mid-October shows the pattern clearly:

That disrupts the usual boom in spending on goods (ie gifts) around the holidays. Instead of limiting spending to an end-of-year holiday shopping spree, Americans are buying more stuff all year, and then spending the same amount on gifts that they used to.

Because of this, GS argues, the Commerce Department’s seasonal adjustments overcorrect at the end of the year, anticipating a boom in holiday shopping that isn’t as extreme as it was before Covid-19. That would, in effect, make holiday retail sales look worse than they are.

This apparently happened last year, according to the bank’s Nov 26 note:

The GS economists also argue that there could be problems with the way the US Bureau of Labor Statistics has worked the pandemic’s sharp swings in employment into seasonal adjustments.

The bank argues that seasonal patterns in continuing jobless claims data show that a similar overcorrection could be happening there, too. Because jobless claims rose so precipitously in March 2020, the BLS’s seasonal adjustments may be overcorrecting as well. This would make labour markets look better than reality between March and September, and worse than reality between September and March:

The bank explains:

These revisions could in principle simply be the result of properly evolving seasonal factors. “True” seasonal patterns often change over time; when they do, seasonal factors should change with them. For example, there has been a gradual shift in not-seasonally-adjusted retail sales from December to October and November over many years as consumers started holiday shopping earlier. The evolution in the seasonal factors for retail sales has taken that shift into account, as it ought to, and the resulting change in seasonal adjustment was therefore not the result of residual seasonality.

However, naturally-evolving seasonal changes are unlikely to explain the predictable seasonal patterns introduced by the revisions shown in Exhibit 2. As evidence of residual seasonality, Exhibit 3 compares the revisions to pre-pandemic seasonally-adjusted continuing claims[1] to the latest vintages of seasonally-adjusted continuing claims in 2023. The high degree of correlation shown (0.45 on a w/w basis, 0.80 on a 4w change basis) indicates that the changes to seasonal factors that arose from the pandemic are likely continuing to have undue influence on this year’s seasonally-adjusted data. Exhibit 4 shows that if taken at face value, this exercise suggests that essentially all of the 182k increase in continuing claims over the last couple of months is attributable to residual seasonality.

Now, GS top economist Jan Hatzius has been very public with his view that a soft landing is in process, and the prior data distortions point in a direction that’s helpful for backing up that call. If holiday retail sales are higher than reported, and continuing jobless claims are in truth lower than reported, it becomes marginally easier to argue that the US economy is doing well, and that Americans needn’t worry about an economic slowdown.

But before readers get too sceptical, the bank also flagged a decline in reliability in Job Openings and Labor Turnover Survey data early this year, which would have made the labour market look better than it was at the time. That was when the Beveridge Curve was all the rage.

GS’s economists also point out that similar issues in initial jobless claims data improved after the BLS corrected them in April of this year. The only issue is that those adjustments probably worsened the seasonal issues in continuing claims, they say.

The good news is that the seasonal distortions in data are on the right track, and are getting smaller as the pandemic recedes into history.

There is a problem that isn’t getting better, however. All types of economic surveys are getting lower response rates, and that trend has sped up since the pandemic:

Keeping that in mind, it’s surely helpful to have professional economists prodding at the quality of data when various data points are important to investors (and policymakers) — however convenient for their house view the conclusions may be.