Premiums Up, or Premiums Down? It Kind of Depends

In the comments to my earlier post on the CBO's premium estimates, reader Janice Doe asks:

I'm confused. McArdle says the CBO says that premiums are going up for those in the individual market. Krugman and Yglesias say the CBO says premiums are going down for the same market (and Krugman also makes a snide comment that Republicans will misunderstand the numbers and claim the CBO says premiums are going up.)

Well I'm a lawyer, so I sure don't understand math, and I most definitely don't understand this mean, modal, median stuff. So can some explain to me in layperson's terms (1) why Yglesias/Krugman and McArdle are reading the CBO differently and (2) who is right (or are both right, but from different perspectives, i.e., mean v. modal v. median)? Many thanks.

Krugman and Yglesias are referring to this report from Jonathan Gruber.  I am referring to the CBO report.  Here is what the CBOsays:

CBO and JCT estimate that the average premium per person covered (including dependents) for new nongroup policies would be about 10 percent to 13 percent higher in 2016 than the average premium for nongroup coverage in that same year under current law. About half of those enrollees would receive government subsidies that would reduce their costs well below the premiums that would be charged for such policies under current law. . . .

Average premiums per policy in the nongroup market in 2016 would be roughly $5,800 for single policies and $15,200 for family policies under the proposal, compared with roughly $5,500 for single policies and $13,100 for family policies under current law.4 The weighted average of the differences in those amounts equals the change of 10 percent to 13 percent in the average premium per person summarized above, but the percentage increase in the average premium per policy for family policies is larger and that for single policies is smaller because the average number of people covered per family policy is estimated to increase under the proposal. The effects on the premiums paid by some individuals and families could vary significantly from the average effects on premiums.

The CBO arrives at this result by assuming that people greatly expand their coverage, but that the cost of delivering that coverage falls due to the efficiencies/rules of the exchange, and the mix of people insured by the exchange gets somewhat healthier.

Jonathan Gruber has a different methodology.  He looks at the projected premiums for the "Silver Plan" (70% of actuarial value), and projects them to a plan that covers 60% of actuarial value using simple arithmetic.  (The "Gruber Microsimulation Model" cited in his charts seems to be a slight misnomer; I can reproduce it in excel in about a minute flat.) A 70% of actuarial value plan costs $5200, so Gruber assumes that 100% of actuarial value costs $7,425 and that 60% of that would cost $4460.  Since the plan that the CBO assumes people will buy under current law covers 60% of actuarial value, he views this as a cost decrease. 

That methodology seems a little off to me.  Either price of a plan with a higher actuarial value should be lower per percent of actuarial value, since providing insurance has some fixed costs; or that ratio should rise, because lowering cost sharing often increases utilization.  Either way, even with a mix of the two effects, it seems unlikely to stay exactly flat.  But I will defer to Professor Gruber on this question.

Which is right?  As I said in the comments, one way to think about it is to compare it with catalytic converters, airbags, anti-lock brakes, and so forth.  If we mandate that everyone get them, the cost of new cars will rise.  But because of economies of scale, the price of new cars might rise by less than it would cost to add these things as options on an individual car. 

So did the price of new cars go up?  I'd say it did.  But you can also correctly point out that now everyone has antilock brakes and airbags, which are valuable things; it's not as if the price of the new car just went up with no added benefit.  Even so, I'm not sure how much sense it makes to do a straight extrapolation back to the old price of cars pre-mandate, and thereby claim that you have actually made cars cheaper.

Obviously, the question gets further complicated if you subsidize 10% of new cars.

To some extent, I'm not sure how much the distinction matters, because this is all absurdly precise.  The CBO is a very valuable institution, but its strength is consistency, not accuracy--it allows you to compare policies, but it does not allow you to see the future.  These are rough estimates, and the true numbers could be either higher or lower.  This is not to slam the CBO, of which I am inordinately fond--they are the first to state all the uncertainties surrounding their estimates.

I think maybe it's more useful to pull away from the numbers a little bit, and look at the various upward and downward pressures that the CBO discusses.

Expanding coverage doesn't just raise premiums because the insurance company is paying for more services, and/or paying a higher percentage of the cost of those services.  It also increases utilization, and not necessarily in ways that increase health.  The CBO expects most of the change in coverage to come from lowering copayments and deductibles, not extra services, so this is a real concern. 

Is that expectation valid?  I am not sure that this was the case in Massachussetts, which is our closest model--but we don't have the same demographic mix as Massachussetts, and our legislation is different in some respects.  With no real way to assess this, I'm going to defer to the CBO and assume arguendo that they are right about decreased cost-sharing.

If we do go along with this assumption, that's certainly going to push costs up.  I don't know if the CBO estimate of a roughly 30% increase in premiums due to expanded coverage is at all accurate, but I think we can safely say that decreasing cost sharing acts to increase premiums.

The CBO also estimates that there will be substantial savings because of enhanced efficiency in the individual market:  about 7-10%.  Again, the direction is almost undoubtedly correct, because administrative costs for providing individual policies are high, and in many states the market is rife with inefficiencies.  The amount?  I think hard to say, but their guess is better than mine.

Finally, the CBO assumes that the mix of people in the individual market will get healthier.  There are two effects here, working at cross purposes.  First, young adults are more likely to be insured than any other age group, and they rarely have expensive chronic conditions.  They make up about 30% of the pool, according to Kaiser.  On the other hand, the uninsured are twice as likely to be in poor health as people with private insurance (11% vs. 5%).  Given the percentages, the effect of the young adults should dominate, especially if a disproportionate number of the very ill get kicked into Medicaid.  Unless the individual mandate fails--as it might, if the penalties are not high enough--this effect should be significant.

Though as a side note, talking about changing the pool mix is somewhat tricky.  Many young adults are absolutely right that insurance is a bad deal for them, particularly in states with guaranteed issue and community rating; they are better off with catastrophic or high deductible coverage.  Bringing those people into the pool lowers the cost for people who already have insurance, but it raises the expenditure of the newer, healthier people considerably.  Measuring the change in the average premium is not the same thing as measuring the change in aggregate expenditure, but people often talk as if they were interchangeable.  For a lot of people, this plan means adding a large new expense which delivers less value in benefits than they pay.

In the end, I'm probably predisposed to assume that premiums will go up.  But the CBO's logic seems pretty sound.  The pool change is significant, but the exchanges will also take on a lot of people who can't currently buy insurance because they have some expensive condition.  The administrative efficiencies are undoubtedly real, but there is simply a limit to how much premium reduction even strong reform can deliver on this front.  If the CBO is right that people significantly reduce their cost-sharing, I'd expect both price and utilization to rise pretty steeply.  In the end, most health care costs are driven by utilization, not administrative costs or outrageous profiteering, so I'd expect that effect to dominate the other two.

But as to the size of the effect, I don't know.  The CBO doesn't either.  In the end, the only way to find out is by enacting this thing.

The only thing I can say with real confidence is that Paul Krugman's interpretation of the whole affair is very odd:

But here's the thing: senior Republican politicians suffer from reading comprehension. (To be fair, the CBO report is written in a remarkably elliptical style). Several have already claimed that the report shows that premiums will rise.

And they probably won't get called on it.
Republican politicians are saying this because this is, in fact, what the CBO report says:  average premiums will be higher.  People may be getting more value for their money.  But there's still more money leaving their wallet every month.  The CBO may be wrong in assuming coverage expansion, but if you go along with their assumption, you end up with higher premiums.  And if you don't go along with their assumption, you have to explain why you reject this assumption, but not all the others.  Neither Krugman nor Gruber really does so.