• Paying marginal cost is good except …

    Inspired by a comment, I thought I should remind folks that paying marginal cost for something is good (and is the price in a perfectly competitive market), except in one circumstance.

    Suppose you don’t want the product. Or, in economics terms, suppose the consumer surplus (personal value) you’d get from it is less than marginal cost or whatever the going price is. In that case, you shouldn’t buy the thing, even at marginal cost (ignore an arbitrage opportunity).

    Turn to Medicare. Suppose Medicare’s prices were marginal costs.* Does that imply Medicare is doing the best it can? That depends what it is buying and for whom, and what value those individuals obtain for what they get.

    No doubt, Medicare could get very good prices for bloodletting services. Should Medicare buy them? What about something else with arguably more health value? How much more health value per dollar is necessary to make it a “good deal”? Lest you think this is devoid of practical application, I encourage you to consider the content of David Leonhardt’s 2009 column on prostate cancer treatment.

    *This is not likely as it would mean the government can determine prices as they’d be set by a perfectly competitive market. Who believes that?

    • @Austin- Thanks for the thoughtful response. Yes, maybe we shouldn’t be paying anything for “bad” care that isn’t worth its cost. I’m not sure we should give bonuses for “good” care, however.

      I think it is still important to remember that the optimal pricing scheme depends heavily on the providers’ goals and their responses to incentives.

      Imagine an “honest” doctor, whose utility = patient benefit – cost of service. He only cares about patient benefit relative to the costs he faces, but has to leave the profession if reimbursement is less than cost. This doc would never prescribe bloodletting when it was reimbursed at cost, since the patient benefit would be lower than the cost. Instead, he would choose a therapy with higher net benefit.

      When marginal cost is constant, even a “dishonest” doctor, who only cares about profit, and whose utility = reimbursement-cost, would not prescribe bloodletting when the reimbursement was set at marginal cost, since profit would equal zero (assuming constant marginal cost).

      And when doctors are somewhere in between, and care about both patient benefit and profit, paying based on value could lead to over-provision of what the government deems a good service (too much of a good thing). A doc might think “When this service was reimbursed at cost, I shirked a little and didn’t provide enough of the good service [we’re back to increasing MC]. I undervalued patient benefit (say by 50%). But now I’m getting paid an amount equal to the full value of the service to the patient on top of the 50% of the clinical benefit I already cared about. Now I’m overvaluing the service at 150% of patient benefit, and I will prescribe even more care than the efficient level.”

      In sum, pricing is complicated! It’s complicated even when we ignore the additional administrative hurdle of government finding out the marginal benefit of the service in addition to the marginal cost. Saying “we should pay for value” is too simple, I think. Ellis and McGuire (1990) JHE gave me a good start for how to think about these issues, though they cover demand side incentives as well as supply side ones.

      • @Aaron – I’m not sure this “honest doctor” stuff is relevant. Did you read the Leonhardt and Orszag ideas I linked to in the last post and Leonhardt’s column I linked to in this one (on prostate cancer treatments? If not, please do. What’s wrong with those ideas?

        If you’ve got three treatments that are equally effective based on credible studies (as judged by some panel, some process) and one costs $5k, one $10k, and one $25k, why doesn’t Medicare just pay $5k for all and let beneficiaries who want the more expensive, but no more effective, treatments pay the marginal cost (or buy additional coverage that would)? The doctors providing them are all “honest”. The costs could all be as low as they can be given the inputs. That doesn’t mean Medicare should pay the full cost of all of them, but it does now. (There’s more to the Leonhardt/Orszag stuff, but this is a one sentence summary.)

    • @Austin- I agree with your comment entirely. However, I wouldn’t have thought of what you describe here as setting prices based on value in the sense that Reinhardt described in his column. He described, I think, setting the price equal to individuals’ willingness to pay (clinical benefit). In your solution, you are still setting a reimbursement equal to a cost (or at least a charge). I’d call that reference pricing. I think that’s a great idea. We don’t want doctors indifferent to three equally beneficial options that cost different amounts.

      I think there has been a lot of thoughtful writing on the theory of payment policy, from papers to books, like Pricing the Priceless. I wouldn’t want people to think that setting reimbursements equal to a maximum willingness to pay value would necessarily result in efficiency, especially since clinical trials would make it hard to find the WTP for the marginal patient, which is the number of interest. Thank goodness some highly effective treatments (like penicillin) are cheap to produce and priced on cost, or things would get very expensive!

      • @Aaron – You’re right. What I described and what Uwe had in mind are different. It’s how I started the post. I know what he means and agree with his implication that it’s not going to happen. I used that as a segue to a reference pricing scheme that incorporates some notion of value, albeit a different one. Pricing the Priceless is on my reading list. (It’s a little embarrassing that I haven’t read it and only recently read some other classics. Better late than never.)