• Understanding the Employer Based Insurance Tax Subsidy, Part II

    It is not my intention to pile on James Kwak. (Can one blogger really constitute a pile on anyway?) I’m a fan of his blogging and he knows far more than I do about many things. But the employer based health insurance tax subsidy may not be one of them. So, with all due respect to Kwak, here goes …

    I’ve already raised one problem with Kwak’s calculation of how much the employer based insurance tax subsidy is to blame for high U.S. health care costs. By e-mail several economists have communicated to me another issue. Actually, I pointed it out in my comments to Kwak’s post as well as to Tyler Cowen’s that quotes the key passage:

    [T]he exclusion gives the median family a discount of 20%. Only about 60% of people get health insurance through an employer plan, so the average discount across the population is only 12%. Given that the price elasticity of health care is almost certainly a lot less than one (if you double the price, demand won’t fall in half), the overconsumption due to the tax exclusion must be less than 12%.

    We now know that the assumption of a tax subsidy at a 20% rate is far too low. The other issue is that the elasticity of health care is not as far below one in magnitude as one might assume. The reason rests on the difference between intensive and extensive margins. Kwak will know what I mean but the uninitiated will not. So I’ll explain.

    When health insurance (or most anything) is made cheaper, more people buy it. The rate of increase of the insured (or, more generally, in purchase of a good) due to price is known as the extensive margin. As health insurance is made cheaper by the tax subsidy, more people buy coverage. But, as Kwak correctly notes, far less than one percent buy coverage when price falls by one percent (elasticity is relatively low). Fine.

    But something else happens when health insurance is made cheaper. People who buy it buy more of it. That is, even if none of the uninsured buy insurance when the price is lower, those who are already insured buy more generous coverage. That’s the intensive margin. Think using a resource that’s already in use more intensively. Another example is when gas is cheaper you drive more. That’s more intensive use of your vehicle (and the roads, and gas). Some people switch from biking to driving a gas-powered vehicle as well, but that effect is the extensive margin.

    As it turns out, for health insurance, the intensive margin is a bigger deal than the extensive one. According to a 2004 paper by Gruber and Lettau that relates the employer tax subsidy to firm insurance offers and insurance spending,

    [T]here is a moderately sized elasticity of insurance offering with respect to after-tax prices (−0.25), and a larger elasticity of insurance spending (−0.7). … Our simulation results suggest that major tax reform could lead to an enormous reduction in employer-provided health insurance spending.

    As the authors point out, the difference between extensive and intensive margin elasticities means that reduction in the tax subsidy would have a greater impact on reducing insurance spending (which is what we want to bring down) than it would on reducing coverage (which we don’t want to bring down). To be sure, there would still be a reduction in employer offers if the tax subsidy were reduced. But the problems that raises can be mitigated or addressed with other elements of reform, like employer penalties and non-group insurance exchanges.

    As I and many economists have said before, the employer based insurance tax subsidy really is a very big deal. I agree with Kwak that it isn’t the entire reason for high U.S. health costs. But it is a bigger reason than he and others might have thought.

    • “…overconsumption due to the tax exclusion must be less than 12%.”

      Can you elaborate a bit more on what 12% means, given e<1? Not sure I am following here. Is it that price change goes from no discount (absent tax treatment) to 12%, in the aggregate for population, and as a result, demand falls 12%-x%?


      • @Brad F – Kwak’s numbers are wrong. A post on Monday will do the calculation with the right numbers. So as not to keep you in suspense, the equation is:

        %change in health spending = %change in price x elasticity

    • Building on my comments to your Part 1 post, wouldn’t overconsumption depend in part on the medical history and age of an individual?

      As I pointed out earlier, when you combine the effect of the tax exclusion with the community rating cross subsidy, the net subsidy to healthy individuals is close to zero, while for those with a medical history the combined subsidy may be many times the value of the tax exclusion.

      If overconsumption is defined as being relative to an unsubsidized rate of consumption, then shouldn’t virtually all overconsumption be concentrated in individuals with high medical costs?

    • Austin- It sounds so much like a tautology that I almost added an immediate clarification. But I thought it was clear that I meant overconsumption, not just consumption. A compounding effect, in other words.

      Practically speaking, this probably just means that employees who expect to have high medical costs will tend to select the most expensive option their employer offers, with the lowest copays, even if it means a high employee contribution. They might also maximize their FSA contributions.

      In contrast, healthy employees should tend to economize, selecting an employer plan similar to what they would purchase on the open market with after-tax dollars.

      As to your second question, I thought I was clear that I was talking about subsidizing premiums, not payouts. I didn’t think insurance implied subsidizing anybody’s risks.