Are most buyers on ACA exchanges previously uninsured?

ACA signups tracker Charles Gaba, extrapolating from recent New York state data, postulates that 85-plus percent of February signups for Qualified Health Plans (QHPs) on ACA exchanges are people who were previously uninsured.

If this inference is on target, it shouldn't be too surprising -- though it either contradicts or suggest a rapid reversal of trends asserted in prior reports from insurers and McKinsey & Co. We know that over 80% of QHP signups qualify for subsidies. That suggests almost by definition that a substantial percentage of signups would previously have found insurance on the individual market unaffordable. Moreover, we also know that less than 20% of those who complete the application process on the exchanges but do not qualify for subsidies actually sign up for a plan on the exchanges.

As I noted in a prior post, these relatively affluent applicants are likely to buy their insurance off-exchange. In the past week I have spoken to or exchanged email with five people who do not qualify for subsidies and have bought insurance for themselves and their families for 2014.  Of these, one bought a plan on-exchange, as a hedge against the possibility of an income drop that would make him subsidy-eligible before year end. Two others located a plan that met their needs on-exchange but bought it directly from the insurer -- which is now extremely easy, since the insurers can't take medical histories into account.  Two more scoped out the market on-exchange, then used a broker to find off-exchange plans that better suited their needs (it appears that in some states you can't beat the on-exchange market off-exchange, and in some you can).

It seems a likely inference to me that most people who previously bought insurance in the individual market are continuing to buy off-exchange, and that most people buying QHPs were previously uninsured. I don't know how to reconcile these inferences with the prior data indicating that most on-exchange buyers (through the end of December) were previously insured. Perhaps I'm overstating the case; perhaps there's been a shift over time; perhaps the early data was partial and misleading (McKinsey's sample was based on just 389 respondents who had enrolled through the exchanges).  Time will tell.

Update, 2/18, 6:05 p.m.: Charles Gaba adds a post noting that in New York, the percentage of signups who were previously uninsured has been rising steadily, from perhaps 50% through December to 75% in January to likely 90% this month.