Why Did The BLS Leak Initial Claims Data 15 Minutes Early?

Today, at 8:21 am, 9 minutes before the official BLS claims release, we received an update from Stone McCarthy which had a rather starting subject line: "Claims Out Early 361,000 -6,000." The email body essentually said what everone would find out was precisely the case subsequently: "Initial Claims for Unemployment Insurance fell 6K to a level of 361,000 for the week ending August 4th. The prior week's level was revised to 367K (previously 365K). SMRA's estimate anticipated a level of 365K for last week's claims. The Bloomberg survey's median estimate was 370K, and individual estimates ranged from 359K to 385K..."  We noted this at the time. Subsequently, Bloomberg confirmed all of this: "Jobless claims data were available on Labor Dept. website about 15 minutes before scheduled 8:30am release today, Stone & McCarthy Research Associates economic analyst Terry Sheehan says." Now aside from the fact that SMRA should be commended for chacking the DOL.gov website early and getting a critical advance look at today's most important data point, we have some questions:

  • Who broke the embargo and why?
  • Who else had even earlier advance access to the data?
  • Why did the BLS have this glitch?
  • And most importantly, especially in light of the recent scandal about the BLS leaking data selectively early, just how is the BLS policing itself from such erroneous data releases, and if the answer is non-existant, can the BLS announce whether it has preferential "huddle" agreements with anyone when it comes to leaking data as it did today?

Because we personally know at least one or two algos that would have loved to respond to the data had it hit Bloomberg's flashing red headline feed 15 minutes early.