Most economic reports should come with a warning: the data is subject to revision. Today’s glowing number that’s the toast of the punditocracy is tomorrow’s statistical detritus. But if the world is always eager to move on to the next number du jour, and forget yesterday’s news, informed analysts recognize that monitoring the initial estimates of a data series, and tracking its path through time, offers another layer of intelligence. For example, a recent study by the Philadelphia Fed found a “small positive (but statistically significant) association between the revision to job gains and the level of job gains.”
With that in mind, let’s review the trend in revisions for private nonfarm payrolls through the decades, courtesy of the St. Louis Fed’s archival database (ALFRED). Specifically, I’m looking at the monthly revision as defined by last reported number (the final data point in most cases other than for the last several months) less the initially reported estimate. The chart below tracks this monthly difference since the early 1970s.
As you might expect, positive revisions tend to be associated with periods of economic growth. But it’s not a hard-and-fast rule. Sometimes the changes are negative (revisions are lower than the initial estimates) and the economy is expanding, although this condition tends to prevail in the early stages after a recession has ended.
The question these days is whether the recent fade in positive revisions is a sign of trouble for the business cycle? For a better view of what’s been happening lately, the next chart summarizes the last 12 months.
It’s clear that the strong run of positively revised estimates for private payrolls has dried up lately. The average revision for the six months through July (we don’t yet know how August’s initial estimate will change) is a slim +21,000. By comparison, the six-month average through February 2012 was a dramatically higher +219,000.
How should we interpret the recent stumble in revisions? History suggests trouble ahead for the economy if the monthly updates turn significantly lower than the initial estimates in the months ahead. That alone wouldn’t be enough to generate a reliable recession warning, but it wouldn’t help either.
You still can’t tell much from cherry-picking the data in search of deeper insight about the broad trend in the economy. Context and perspective, by analyzing a broad spectrum of indicators, is essential. In other words, darker signals from payrolls revisions may, or may not be, a smoking gun. Much depends on how the numbers look overall. On that note, the trend signals from most other key indicators remain positive, based on the latest data through July. August is a mixed bag so far, but we only have a few numbers at this point. But if a tipping point awaits in the near future, it’ll be obvious in the numbers, including revisions to payrolls.
Correction: Previous versions of the two charts above included titles that reversed the order of the differencing between revised and initial data for employment. The corrected charts now include titles that reflect the data: revised payrolls estimates less initially reported data. Sorry for the confusion.
Good article and great data charts