The latest “Economic Beat” column from Barron’s Gene Epstein highlights an interesting quirk in government reporting of month-to-month economic statistics, offering a reminder that it’s important to focus more on trends than the latest numbers.

While the Bureau of Labor Statistics reported no increase at all in average hourly earnings in June (after a string of gains through the first five months of this year), some analysts point out that the downside surprise had a lot to do with the fact that June 12 landed on a Friday.

Call it Friday-the-12th fatigue. It turns out that when the 12th is a Friday in any reporting month, average hourly earnings suffer a fainting spell. That last happened in June. But given strong fundamentals, a rebound in average hourly earnings is probably happening this month.

Why the fainting spell? Data on average hourly earnings are reported in the BLS Establishment Survey, and as the Technical Note that accompanies these employment releases explains, “In the establishment survey, the reference period is the pay period including the 12th, which may or may not correspond directly to the calendar week.”

In other words, employers who respond to the survey are instructed to record earnings in a pay period that includes the 12th. But in this case, the 12th was on Friday, the end of the calendar week. And it’s quite possible that checks paid bimonthly—say, on the 14th or 15th—were not reported in the survey questionnaire, an omission that would not have occurred had the 12th landed on a Tuesday. The non-reporting of those checks might have caused a downside distortion.

I tested this theory with the aid of research assistant Gabriel Alpert. We took the past 32 cases in which the 12th landed on a Friday, and where in the previous and following months, the 12th did not land on a Friday, an exercise that takes us back to August 1994. To make sure that I was examining “real time” data, we looked at the original releases rather than at data that might have been revised.

The pattern took on plausibility as we worked backward in time. In December 2014, September 2014, and July 2013—all months with the 12th landing on a Friday—average earnings were flat or declined, with increases in the month before and in the month following.

In all, 25 out of the 32 cases—nearly four of five—revealed that same basic pattern: Friday-the-12th months showed weakness compared with the month before and the following.

So the Friday-the-12th-fatigue theory holds up. And if it still applies, expect a rebound in average earnings for July (to be reported on Aug. 7), when the 12th falls on a Sunday.