Spanish has had fewer *visitors* this year...but "Visitors" don't necessarily correlate to "Enrollees"

The Washington Post's Amy Goldstein posted an interesting story the other day regarding the 3rd Weekly Snapshot report, focusing on the Hispanic enrollment issue:

The number of people shopping for medical insurance on the Spanish-language version of is lagging behind last year's interest, even as the Obama administration urges Hispanics to sign up for coverage under the Affordable Care Act.

Federal figures released Wednesday show that about 153,000 people used during the first three weeks of the current enrollment season for ACA health plans, down from 244,000 during the same period a year ago.

The enrollment snapshot for the week ending Nov. 21 also indicates that the federal insurance exchange continues to have greater difficulty in attracting new customers than last fall. Overall, the 1.6 million people who have chosen new health plans so far exceeds the comparable number from 2014. But 35 percent of them are newcomers, down from 48 percent at this point a year ago. That equates to 576,000 newcomers who have selected health plans – nearly 90,000 fewer than last year.

I already wrote up a piece regarding the 3rd paragraph above; this is the same issue which Jed Graham wrote about over at Investor's Business Daily last week. As for the issue of Hispanic enrollments, let's take a look at the numbers in question:

Sure enough, that's a whopping 37% drop in the number of people who used, year over year.

The first thing I wanted to look at was how many people are using the English version of the site, HealthCare.Gov itself:

  • 2015: 7,942,195
  • 2016: 7,288,389

That's an 8% drop year over year, which says that total "users" of both websites are down about 9% for the first 3 weeks. The Spanish version is still down to a significantly higher degree, but that does explain at least some of the discrepancy. Yet actual QHP (qualified health plan) selections are up nearly 20% so far. What's the deal?

Well, the main thing you have to remember is this: Users actually means unique visitors to the websites, not the number of people who submit an application or actually select a plan: or Users: These user metrics total how many unique users viewed or interacted with or , respectively, over the course of a specific date range. For cumulative totals, a separate report is run for the entire Open Enrollment period to minimize users being counted more than once during that longer range of time and to provide a more accurate estimate of unique users. Depending on an individual’s browser settings and browsing habits, a visitor may be counted as a unique user more than once.

This may sound like picking nits, but this is a hugely important distinction. Last year, during the first 3 weeks, 1.38 million people selected policies out of 8.18 million unique visitors total (both sites combined). That's around 5.9 site visits per plan selection, not broken out between the two. This year, there were 1.65 million plans selected out of 7.44 million unique vistors, or 4.5 visits per selection.

What does this tell us? Well, frankly, to me it just means that people are becoming more familiar with how the websites and the process works, now that the bugs are mostly worked out. Remember, the 2014 experience was a disaster for most people, especially in the first ugly few months. For 2015 they completely changed the interface and procedure for most of the process, but I'm guessing most people were still pretty gun shy and probably kicked the tires more before committing, visiting the site several times before actually enrolling. Also bear in mind that a lot of those visits were made by people with no intent of actually enrolling anyway--reporters, tech geeks, insurance company folks, politicans and their staff.

This year, while there have been further enhancements/improvements, I'm guessing it's starting to become old hat to everyone involved. Journalists, politicos, tech writers and so on aren't nearly as interested in what's up with HealthCare.Gov; it's just not nearly as interesting to most people. Furthermore, even those enrolling are going to be less timid; instead of poking around 3-4 times, they might only do so twice (once for window shopping, once to actually enroll), and so on.

Frankly, I expect the "visits to enrollment" ratio to shrink every year as more and more people familiarize themselves with the site. Think about the last time you bought something on Amazon, after probably just loaded the site, put stuff in your cart and checked out. In 2015 it was a 6:1 ratio. In 2016 it's 4.5:1. In another 5 years it could be down to 2:1 or so for all I know.

Of course, by the end of the 2015 Open Enrollment Period, this ratio had dropped substantially: 35.2 million total unique visits for 8.84 million plan selections, or around 4.0:1, so who knows? Another data point: By the end of #OE2, 11% of those who self-reported their ethnicity listed "Latino", but I'm not sure how useful that is since over 36% of the total enrollees didn't report an ethnicity at all.

OK, so the "Visitors vs. Enrollees" factor would explain the overall reduction, but why the discrepancy between the English and Spanish sites?

Well, there's two additional factors I can think of which might be relevant.

First, this year's Open Enrollment Period started 2 weeks earlier than last year...but the deadline for January 1st coverage remains the same (12/15). I don't know anything about differences in cultural factors, but I suppose it's possible that the Hispanic community in general has a tendency to wait until deadlines are closer to sign up than the non-Hispanic community. That's not a slam, just speculation, and If I'm wrong about this I'm sure I'll be roundly torn to shreds in the comments section. Related: It's worth noting that last year, Week 3 included Thanksgiving, whereas this year Thanksgiving won't be added to the tally until Week 4's report; for all I know, that makes some sort of difference, although probably not.

The other possible factor is one which I hesitate to bring up, but what the hell. Consider this:

  • According to Wikipedia, as of 2006, around 80% of undocumented immigrants to the U.S. were from Mexico or another Latin/Central American country. I'm assuming that the vast majority of these folks speak Spanish, but please forgive my ignorance if I'm wrong about this
  • According to CMS, as of June 30th, around 423,000 effectuated enrollees had been dropped (involuntarily) from their policies specifically due to being unable to properly document their legal residency status in the United States.

Add these three factors together and I have a half-baked theory which, again, could be dead wrong:

I'm assuming that has substantially improved their residency verification system since last year; presumably they've added new features which are designed to weed out undocumented immigrants and/or others with legal residency issues up front, which hopefully reduces the number of folks who manage to select and enroll in a plan for a few months before being kicked off of it.

I have to wonder whether part of the reason for the seeming reduction in Spanish-language website traffic (despite HHS pushing hard this year to increase Hispanic enrollment) might be because a big chunk of that market is visiting, attempting to enroll but then are being blocked due to residency verification issues up front, and thus aren't bothering to come back?

IF this theory is correct, it should mean that initial enrollment numbers may be lower because of it, but attrition levels should also be lower. In other words, the initial number may be 400K or so lower, but the attrition number later on should also be 400K fewer.

UPDATE: Here's another way of looking at the "legal residency" issue:

  • Let's assume that the 11% of enrollees who self-identified as "Latino" is representative of all 8.84 million who selected plans via during OE2 (970K).
  • Let's also assume that at least 75% of the 423K who were kicked off are exclusively/primarily Spanish-speaking (based on the 80%-Mexican/Latin/Central American factor noted above)

  • If so, that means that fully 1/3 (317K out of 970K) of all Spanish-speaking enrollees on were kicked off their plans versus just 13.5% of English-speaking enrollees (106K out of 7.87 million)

If this is accurate, it explains a lot...because you also have to remember that for the first 3 weeks, 65% of all plan selections were renewals by current enrollees. Again, if accurate, then only around 7.7% of current enrollees are likely to use the Spanish version of the website, even though 11% of last year's enrollees were "Latino".

Anyway, it's a theory. I apologize if I'm displaying cultural bias in presenting it, but it seemed worth at least throwing out there.

Of course, all of the above could also be a big crock of hooey; it's entirely possible that I'm wrong about all of this, and that the ACA is having a serious problem with Hispanic enrollment this year.

Oh, one more thing: Again, just as "Unique Visitors" doesn't equal "Plans Selected", it also doesn't equal "Applications Submitted". When you compare those year over year, you get a very different picture (unfortunately, this isn't broken out by English/Spanish):

  • 2015 (as of Week 3): 2.53 million apps submitted
  • 2016 (as of Week 3): 2.91 million apps submitted (15% more)

Finally, unfortunately, Covered California hasn't released any of their data on renewed policies as of yet; seeing how California is 1) the largest state in the country and 2) also has a huge Hispanic population, it's hard to draw any national conclusions about the topic without their data being included.

Anyway, the Fourth weekly report should be released on Wednesday. While I'm expecting the weekly enrollment number to be down about 30% since last week due to Thanksgiving, it should still bring the cumulative total to just over 2.0 million, and it will be interesting to see how the Spanish-language numbers fare.