Designing a High Deductible Plan

I am increasingly gratified to see more economists and politicians coming out in favor of high deductible health insurance. The positive effects of such plans would be both direct and indirect. I am actually more excited by some of the indirect, subtle effects that I think would happen as more people moved into high deductible plans, especially in regard to demanding more price information.

Here are my thoughts on how one would go about thinking about designing a high deductible plan for a self-insuring employer. There are lots of details that I won't get into, and most important, to make real progress I would need historical data on the distribution of health expenses in the employee base. But I think I can illustrate some of the key ideas.

The data that I would start with would be the percentage of employees with yearly expenses falling in different ranges, like this:

Health Care Expenses, % of Employees

0 - 2500, 10%

2500- 5000, 15%

5000 - 7500, 25%

7500 - 10000, 20%

10000 - 12500, 10%

Greater than 12500, 20%

If I had this data for an employer, I would start with it to get a feel for where a reasonable initial deductible might be. I want a deductible high enough so that I capture a reasonable number of employees with total expenses under that amount. At the same time, I don't want a deductible that is going to be unreasonably high. What's reasonable? Well, for the number of employees, I think we would want to catch something like 25%-50% of the base with total expenses under the deductible, at least. My thinking here is that the deductible is set to capture expenses for which true insurance makes sense, and if something is occurring more than 75% of the time, or even up to 50% of the time, it sounds too common to be reasonably covered by insurance. But on the other hand, I don't think a deductible that is too high is going to be acceptable to people who are used to "insurance" paying the bills.

Since I don't have that detailed of data at hand right now, let me proceed by assuming that what I do know is that 1/2 of the employees have total yearly expenses under $7500, and that the overall average expenses are $18,000 per year. For a family, that is a reasonable number. Given these two assumptions, we can infer that the other 1/2 of the people have expenses that are on average $32,250 per year (with some no doubt having very high expenses!)

With average expenses of $18,000 per year over all employees, we know that "fair" insurance would be priced at $18,000. (Apologies again to Sec. Sebelius for using such a blasphemous phrase as "fair insurance. What I mean by fair here is just that if the employer charged $18,000 for the insurance, with no deductible, it would come out, on average, just even.) I do recognize that the data we observe will be influenced by the deductible in place during the data collection period.

Now what would happen if we put a $7500 deductible in place? Well, everyone with expenses less than $7500 would pay all their health expenses themselves. That is 1/2 of the people. The other half would pay their deductible, and the employer would pick up the rest.

If expenses in the upper half of the distribution (greater than $7500) stayed the same -- I will return to this point -- then the employer's expected expenses look like this:

Employer's Expected Expense = .5(0) + .5($32,250 - $7500)
= $12,375

That would be the "fair" price of the insurance plan with a $7500 deductible. Note that the fair price of the plan with a deductible is not just the average expenses less the deductible -- that is, the fair price is not $18,000 less $7500 = $11,500.

So, there is our comparison: Insurance that covers everything for $18,000, or a $7500 deductible policy that would cost only $12,375.

Importantly, as a consumer, I could buy the cheaper policy and put the difference in prices, $18,000- $12,375 = $5625 into an account, which we might as well call a Medical Savings Account. On average, that amount of money will cover my out-of-pocket medical expenses (1/2 of the time my expenses will be less than $7500, or $3750 on average; and half of the time my expenses will be the deductible, $7500.) I am not dealing with taxes here, but if that MSA better be tax deductible if the cost of insurance is, or this will never work. Also, the MSA cannot be "use it or lose it."

That gives an idea of some of the thinking that would go into the design of high deductible plan. Next would come some more subtle, yet important, issues. One, what would happen to the expenses of those folks who used to have expenses greater than $7500. I would expect them to come down, for several reasons: One, the employees would simply not incur as many expenses, partly because they would decide to forego some expensive but optional services. Two, because they would be more careful about their health to begin with. Three, because they would put some pressure on health care providers to cut their prices. These effects would be the cost-control measures that we so desperately need, and they would allow for a DECREASE in the price of insurance as time went on. Can you imagine that??

Another subtle issue would be self-selection if we made two plans, the high deductible and the no deductible, both available. Then we would get folks taking the no deductible plan who expected to incur large expenses, and vice versa for the low deductible plan. That would allow the high deductible plan to be priced even lower, and would force a higher price on the no deductible plan. This is essentially what we see happening to individual insurance prices in the California market, and it should not be viewed as a bad thing.