“When you look at the 6 mil­lion Amer­i­cans who have lost their poli­cies and some — they claim 4.2 mil­lion peo­ple who have signed up — I don’t know how many have actu­al­ly paid for it — that would indi­cate to me a net loss of peo­ple with health insur­ance.”  House Speak­er John Boehn­er (R‑OH)

Boehner’s state­ment last month that few­er peo­ple had gained cov­er­age through health exchanges than had lost their insur­ance due to can­ce­la­tions required by the health care law was met with wide­spread deri­sion. The state­ment earned him 4 Pinoc­chios from the Wash­ing­ton Post’s Fact-Checker.

But a new sur­vey released by the RAND Cor­po­ra­tion at least par­tial­ly vin­di­cates the Speaker’s claim.

The Admin­is­tra­tion has for weeks been tout­ing the survey’s topline esti­mate that the num­ber of unin­sured peo­ple had dropped on net by 9.3 mil­lion since last fall. RAND, how­ev­er, did not release the sur­vey until yes­ter­day, so it was impos­si­ble to deter­mine just how it had arrived at that figure.

Now we know. Between last Sep­tem­ber and mid-March, accord­ing to RAND, 5.2 mil­lion peo­ple had lost their health insur­ance cov­er­age, while only 1.4 mil­lion unin­sured peo­ple had obtained cov­er­age through the exchanges. Those fig­ures are very close to those used by the Speaker.

So why did RAND con­clude that there has been a net gain of 9.3 mil­lion new­ly insured between Sep­tem­ber 2013 and mid-March 2014? The biggest rea­son is some­what sur­pris­ing.  The sur­vey found that 8.2 mil­lion addi­tion­al peo­ple have enrolled in employ­er-spon­sored cov­er­age since last September.

This num­ber seems rather high, par­tic­u­lar­ly since the Admin­is­tra­tion decid­ed not to impose a tax penal­ty on busi­ness­es that don’t offer cov­er­age to their work­ers this year. We do know that mil­lions of peo­ple have in the past declined insur­ance offered by their employ­ers.  It may be that many of them decid­ed to accept it this year, per­haps to avoid the tax penal­ty that the IRS will assess on unin­sured indi­vid­u­als. And more peo­ple under 26 may have glommed onto poli­cies offered through their par­ents’ employers.

The oth­er major source of new cov­er­age was Med­ic­aid.  Accord­ing to the sur­vey, enroll­ment in the pro­gram increased by 5.9 mil­lion. But it is impos­si­ble to deter­mine how many of these new recip­i­ents were eli­gi­ble for Med­ic­aid under rules that pre-exist­ed Oba­macare. RAND, for exam­ple, does not pro­vide details of how many of these 5.9 mil­lion reside in the 25 states that did not expand eli­gi­bil­i­ty or how many of those liv­ing in expan­sion states would have been able to enroll under the old rules.

The RAND sur­vey is but one data point among many in assess­ing enroll­ment under Oba­macare. Since it is a sur­vey, it is sub­ject to con­sid­er­able uncer­tain­ty, as Avik Roy has not­ed.  And since it con­clud­ed in mid-March, it does not account for enroll­ments that occurred dur­ing the final days of open sea­son.  Its find­ing that employ­er-spon­sored cov­er­age has spiked is hard to square with a recent ADP study, which report­ed that the num­ber of young peo­ple enrolling in employ­er-spon­sored cov­er­age is in decline.

But if the results bear up, then the exchanges could prove to be some­thing of a dis­ap­point­ment, with the num­ber of new­ly insured peo­ple falling short of the num­ber of can­ce­la­tions.  The Med­ic­aid expan­sion, too, will have been less effec­tive than imag­ined, par­tic­u­lar­ly giv­en the dif­fi­cul­ty of deter­min­ing how many new enrollees were eli­gi­ble under the old rules.  That would leave employ­er-spon­sored cov­er­age as the biggest fac­tor in reduc­ing the num­ber of uninsured.

It would be iron­ic indeed if the major effect of Oba­macare was to induce mil­lions of peo­ple to sign up for an employ­er-spon­sored plan that was avail­able to them all along.