Major QHP discrepancies between HHS & Exchanges for 6 states UPDATE: (sigh) OK, I've figured out some of them...

I know I said I would save any further analysis of today's monthly HHS report until tomorrow, but when I started plugging in the state-by-state numbers, I came up with some really strange discrepancies between the QHP selection number in the report and the corresponding report from some of the state exchanges for the same date.

It got to be so strange that I actually whipped up a little spreadsheet for all of the state-based exchanges (minus Massachusetts, which is a special case anyway; plus Oregon, which is running off of HC.gov this year but has a large discrepancy from the official State of Oregon Website which is reporting enrollments this year as well). The yellow lines are the states I'm most concerned about (the others all have either obvious explanations or are minor enough not to worry about). Afterwards I'll run through the problems with each one.

• California: Not only does the HHS report not include any renewals (either automatic or active/manual), which means there's likely up to 450K manual enrollments missing from the report (plus another 550K autorenewals missing since then), but it also leaves out the biggest day of the period so far...December 15th! This is noted in the HHS report and explains the missing 25,408 enrollees, but that's not exactly a minor point!

• Colorado: When I saw the whopping 108K figure reported from Connect for Health Colorado through 12/15, I naturally assumed that this included both manual & autorenewals. There is a reference to several other states having autorenewals included in the 12/15 HHS report, but Colorado isn't among them, so I'm not sure what the deal is here. Still, the difference is so massive that it's pretty obvious that CO's 108K figure does indeed include autorenewals, so I'll assume that they just decided to throw them into their own report but not HHS's.

• Connecticut: OK, CT is indeed one of the states which has autorenewals included in the HHS report. Even so, there's an 8,400 discrepancy between HHS's 77,042 and Access Health CT's 85,466 (19,402 new + 66,064 renewals).

• District of Columbia: Both the exchange and HHS numbers refer to new additions only. There's a difference of 183, but the dates line up and the numbers are all pretty small, so I'm going to assume that this is a case of a clerical errors, double-entries or something similar. Also, they had about 500 applications with technical issues, so it sounds like some of them were worked out while others weren't.

• Hawaii: Like DC, the numbers are small, of course, but unlike DC, the difference is much larger and there's no obvious explanation for the discrepancy: The exchange reported 3,500 enrollees, the HHS report says it's only 1,903. On the other hand, the 3,500 figure came from a news report, so perhaps they screwed that up?

• Idaho: The state exchange reported exactly 74,689 as of 12/14. The HHS report claims 73,262 as of 12/15. Not a huge difference, but how could the exchange report be 1,400 higher if it only ran through 1 day earlier??

• Kentucky: Like CA, this one is easy: The exchange number of 85K includes one more day than the HHS number of 82.6K.

• Maryland: The Maryland Health Benefit Exchange reported exactly 61,031 QHP selections as of 12/15. The HHS report says it was only 50,742. That's a difference of over 10,000 people...what gives? I have no idea.

• Minnesota: Again, the exchange number includes 2 extra days.

• New York: The exchange includes 1 extra day; plus, it's possible that some of these are from NY's unique "Child Health Plus" program (not to be confused with CHIP). NY State of Health had a habit of throwing this in with their own reports last year (without mentioning it), which resulted in a 40-60,000 difference between their reports and HHS's; it took awhile to figure out the mystery at the time.

• Oregon: Yes, I know they're running off of HC.gov this year anyway, but the state of Oregon has been running weekly reports on 2015 enrollments (both on and off-exchange) anyway...and something doesn't add up. Just a few hours ago, the Oregon Dept. of Consumer & Business Services reported 67,467 private policy enrollments through Healthcare.Gov between 11/15 - 12/21. They specify that this does not distinguish between paid or unpaid enrollments.
  However, the HHS report claims 73,152 QHP selections between 11/15 - 12/15...6 days earlier. Why is the state site reporting nearly 5,700 fewer from 6 days later?

• Rhode Island: In this case, the number of enrollees reported is identical for both the state exchange and the HHS report. However, the state exchange report was supposedly from 2 days earlier...through 12/13. I find it difficult to believe that not a single person in Rhode Island enrolled over the next 2 days (which was going gangbusters in every other state in the country).

• Vermont: I don't have a 12/15 report from the VT state exchange, but they did release one from 4 days earlier. It claimed 25,867 total QHP selections for 2015. The HHS report says it was 21,709...as of 4 days later. Again, a 4,100 difference.

• Washington State: Like CA, KY, MN & NY, there are extra days included in the higher-numbered report, so this isn't an issue.

The explanation for some of these may be found in the fine print/footnotes for the HHS report (that's how I discovered the first one in California). In other cases it may be a misunderstanding on the part of either myself or a reporter (Hawaii likely falls into this category). In still other cases, it might be simple clerical errors on the part of either the exchange or HHS.

Even so, there seem to be a lot of weird discrepancies going on here. If anyone is able to figure out any/all of the above, please feel free to let me know.

In the meantime, I'm going to stick with the number provided by the state exchanges in all cases on the Spreadsheet (regardless of whether it's higher or lower than HHS's) until further notice. The exception to this is Oregon, where there is not official number provided by the state exchange itself.

UPDATE: ARRRRRRRRGH!! OK, this doesn't explain everything, but it certainly clears up some of the discrepancies; I just noticed this at the top of the Page 16 Appendix Table B1:

See that second line? The HC.gov data runs from 11/15 - 12/15...but the state-based data (with the exception of California) only runs through December 13th, not the 15th.

Those 2 extra days make a huge difference, given that the 14th & 15th were deadline dates (or at least the original deadline dates) for Colorado, Connecticut, DC, Hawaii, Idaho, Kentucky, Minnesota, New York and (I think) Vermont.

In other words, I think I've just cleared up the mystery in at least 8 states all in one shot. The enrollment number given by the state exchange as of 12/15 was higher than that given by HHS because the HHS number only ran through 12/13...2 days less. GROAN...

That still leaves unexplained weirdness in DC, Oregon, Vermont and Washington State. I'm leaving the original chart up above, but have revised it below...