Tag Archives: dplyr

Comparable SHOP policies are more expensive than individual ones: Why?

One of the ideas behind employee group policies and one of the purported virtues of employer sponsored health insurance is that the marketing and other overhead costs are lower.  Instead of selling 200 policies to 200 persons, the insurer sells one policy to an employer.  And one reason employer sponsored health insurance has often been supposed to be lower priced than comparable individual policies is that individuals who are gainfully employed tend not to have at least some of the expensive chronic or acute conditions such as certain forms of cancer or serious heart disease.

These assumptions predict that if one looks at policies sold in the individual exchange and the SHOP exchange under the Affordable Care Act that are sold in the same county, from the same issuer, have the same marketing name, have the same metal level, the same plan type (PPO, HMO, etc.), the same deductible and the same out-of-pocket limit, the SHOP premiums should on balance be lower than the individual premiums.  Apples to apples, they should at least be no higher. Indeed, lower potential premiums are one of the key reasons for the existence of SHOP exchanges.

When I examine the data available from healthcare.gov, however,  I find that the opposite is true.

SHOP policies are on average 8% more expensive than  apparently identical individual policies.

That’s the key finding.  The rest of this post tries to figure out why this might be the case. I’ll tell you the statistical ground I traversed in this effort. But I will confess that at the end of the day I am not quite sure why it is that SHOP premiums sure appear to be higher.

It’s not a matter of outliers skewing the mean.  The median premium ratio between comparable SHOP and individual ratios  — the “SHOP/individual ratio” — is 1.08. 10% of the comparable policies are more than 21% more expensive on the SHOP exchange. And even the lowest 10% are just 2% cheaper on the SHOP exchange than on the individual exchange.

The figure below shows the distribution of SHOP/individual ratios among the 12,689 comparable policies in the dataset. As can readily be seen, most policies are more expensive on the SHOP exchange. (A SHOP/individual ratio greater than 1 means that the median SHOP policy was more expensive than the median comparable individual policy).

nationwidePremiumRatioHistogram

Breaking down the data to gain insight

The state in which the policy was sold and the issuer might affect the SHOP/individual ratio.  Unfortunately the effects of these potentially separate variables are extremely to tease apart because no issuer sold in more than one state.   The table to the left shows the results. It shows significant variation in median SHOP/individual ratios across the combination of issuer and state.   On the top end, Minuteman Health,  which sells in New Hampshire has a ratio of 1.31 and Health Alliance Medical Plans, which sells in Illinois has a ratio of 1.25 Montana Health CO-OP has a ratio of 1.2. On the bottom end, CommunityCare, which sells in Oklahoma has a ratio of 0.76 and CoOpportunity Care, which sells in Iowa, has a ratio of 0.86.

 

StateIssuermedianRatio

The best I can do to tease out whether it is the state in which the policy is being sold or the issuer in the state that is responsible for the variation is to look at those states in which all  the issuers have SHOP/individual ratios greater than the median of 1.08 and none have a lower one.  Two states show up: Montana and Texas.  This suggests some regulatory issue or special market condition in those two states that is leading SHOP premiums to be unusually high or individual premiums to be unusually low.  Unfortunately, the finding is not particularly robust and I will confess to having little idea what this special factor might be.

We can see if the metal level and the plan type end up affecting the SHOP/individual ratio once the issuer and state are controlled for.   Multivariate linear regression shows the impact of Metal Level and Plan Type to be quite small.  The greatest effect was shown by POS plans, which reduced the SHOP/individual ratio by 0.03.  Everything else had a smaller effect. Basically, SHOP plans tend to be more expensive than comparable individual market plans without regard to metal level or plan type.

An adverse selection theory?

Let me at least explore one other reason SHOP premiums might generally be higher.  There are several buffers against adverse selection — the proclivity of persons with accurate knowledge of higher risk to select greater amounts of insurance coverage — in the individual market.  A key one of these are the premium subsidies, which mean that many individuals, even those of relatively low risk, pay less for insurance than their expected cost of health care. There’s the individual mandate that likewise coaxes individuals, even those of low relative risk, to purchase insurance.  There’s an open enrollment period. An individual can’t easily wait until they get sick and then by insurance in the middle of the year.  In theory, they only can purchase insurance outside of the open enrollment window if they qualify for “special enrollment” by virtue of a small set of non-medical changes in their life circumstances.

The SHOP market may be more vulnerable to adverse selection problems.  There’s no employer mandate that applies to small employers that tend to be eligible for SHOP policies. Although there are some tax credit subsidies, they tend to be less lavish than those bestowed on individuals and they apply only to some small employers.  For other small employers there really aren’t any of the sort of “special deals” that might make it a good deal for even those whose employees and dependents are low risk. Finally there is no limited open enrollment period. On the contrary,  under 45 C.F.R. § 155.725(b), “[t]he SHOP must permit a qualified employer to purchase coverage for its small group at any point during the year. ”  Thus, an employer can wait until someone they care about — say the CEO’s daughter — gets really sick and then decide it would be prudent to have a employee group health plan.

What we may be seeing in the SHOP exchanges is an insurance world in which even the mild protections against adverse selection contained on the individual exchanges do not exist.  And the result is predictable: higher prices and very few buyers.

Conclusion

So, what do we do with this finding?  What do we do about the fact that SHOP policies tend to priced higher than comparable individual ones?  Standing by itself, I am not sure one can draw much of an affirmative policy implication out of the result, particularly where we don’t yet have a good handle on causation.  I do think it argues for thinking about imposing some adverse selection controls — such as limited open enrollment periods — if we are going to keep SHOP alive.

I also think, however, that the findings disclosed here have kind of a rebuttal value.  They mean that proponents of SHOP exchanges should have a difficult time grounding an argument to preserve that additional complexity of Obamacare on grounds that it saves money. Although there are, to be sure, exceptions, and although there may be other reasons to induce small employers to purchase health insurance for their employees, right now it does not look as if the price is right.

 

Share Button