Visualizing The BLS' Establishment Survey Revisions
As part of today's non-farm payroll release, the BLS also issued its revision to the Establishment Survey as a result of updated population estimates, which, as the name implies, adjusted monthly data to the Establishment Survey which is the actual headline print that moves the market, not the data in the household survey which is what the unemployment rate is based on. In short: of the 12 monthly revisions, there were just 2 months in which the post-revision data was adjusted downward, July and August, with all other months supposedly adding jobs to the cumulative total, which as of December stood at 134,668 jobs as revised, compared to 134,021 pre-revision. So for those curious how the sequential change in jobs would have looked on a pre vs post-revision basis, we summarize the 2012 data in the chart below. In short: the revision would have added a total 335,000 jobs to the Establishment Survey headlines over the 2012 NFP headlines. The point of the chart is to show just how variable the actual monthly swing is based on exit assumptions, yet this is precisely the data that the kneejerk collocated algos trade on.