Marginal Utility

by Tom Bozzo

It's not too often that I strenuously disagree with Jeremy Freese on such things, but his contrarian post taking exception to Paul Krugman's column from last Friday, which described some important ways in which the Medicare prescription drug benefit sucks as enacted, is such an occasion. Jeremy is especially irritated that Krugman condemns the benefit "doughnut hole" (the range of prescription drug expenses over which Medicare Part D provides no benefit) while failing to note the provision of benefits for 'catastrophic' expenses. Jeremy writes:

Because the doughnut hole can be so easily described in ways that make it seem strange, everybody seems to have fixed on it as though it was some kind of crazed (and, by Krugman's insinuation, deliberate) irrationality. While it's not the way I would designed things, the doughnut hole--and even prohibiting people from buying insurance to fill it--does make sense when you understand that the prescription drug benefit is really two separate benefits stapled together. (*)

That all depends on what the meanings of "strange" and "crazed deliberate irrationality" are.

Honesty requires me to report that the doughnut hole is not purely a Republican concoction. A CBO reporton design parameters for a Medicare drug benefit notes that a Clinton administration plan featured a doughnut hole of its own. This does not pose a consistency problem for Krugman, who criticized "New Democrats" in a prescient column from September 10, 2000 for "penny-ante activism, handing out a few baby carrots here and there to people who do the right thing." Moreover, plans without the doughnut hole had been advanced by Senate Democrats, including an alternative proposal from former Sen. Chuck Robb discussed by the CBO and a related plan that Krugman preferred to the Republican version in a June 18, 2002 column:

The Senate Democratic plan would cost about $500 billion over the next decade; if we could afford that $1.35 trillion tax cut, we can afford prescription drug coverage...

Of course, the House Republican plan, with a price tag of $350 billion, looks even more affordable. What's wrong with it?

One answer is that in order to save that $150 billion, the Republican plan leaves many truly needy retirees behind. The Senate Democratic plan imposes fairly hefty co-payments, but then covers all subsequent expenses. The House Republican plan provides pretty good coverage for the first $1,000 in expenses, much less coverage for the next $1,000, and nothing at all after that until you reach a $4,500 annual limit. So a retiree with, say, $6,000 in drug expenses would find himself paying the full $4,500 — a crippling expense for many families.

The obvious rational basis for the hole was to hold down measured program costs and thus help entice recalcitrant members of Congress to vote for it. Given that its origins are political, the aim of the administration in charge makes a difference. Crazed New Democrats may be inclined to offer weak tea programs on the grounds that some program is better than no program, and popular entitlements will not only be hard to repeal, but also create pressure to remedy the hacks required to pass the original legislation. In contrast, crazed Republicans are ostensibly unfriendly to entitlements (except, in practice, as political tools), and some of them really do operate under the truly crazed theory that bankrupting the government in order to force otherwise popular programs to be scaled back is a good idea.

Is the hole a strange feature, as Krugman asserts? Well, yes. The CBO report on drug benefit design notes that coverage holes are rare in private-sector insurance, though I note that benefit caps are not. The economics of choice under uncertainty imply that it would be unusual for risk-averse consumers to choose to insure small losses but not larger losses. (**) This is to say that as an insurance feature, the hole doesn't makes much sense — or, to be a bit more precise, as Isaac is in the JFW comments, there might be a hypothetical world in which it could make sense, but that happens not to be the one in which we live.

From another perspective, saying that the hole — and the prohibition of Medigap insurance to fill the hole — constitutes good public policy is equivalent to saying that it makes sense to raise the price observed by beneficiaries just as their drug expenses start to get serious though not "catastrophic." Or, the 251st dollar of drug expenses is more socially valuable than the 2,251st. A case for that can be made. It's likely to be counterproductive, for instance, to discourage people from taking medicines for chronic conditions only to see them land in the hospital for them. The catch is that the benefit parameters have been set such that it is not hard to run out of the initial benefit before running out of chronic conditions worth medicating.

But, as Krugman notes, the official justification is even weaker. From the offending column:

According to a report issued along with the final version of the bill, people are prohibited from buying supplemental insurance to cover the doughnut hole to keep beneficiaries from becoming "insensitive to costs" - that is, buying too much medicine because they don't pay the price.

This justification is especially questionable as my understanding of the relevant research is that people don't do great jobs of weighing the true costs and benefits (internal, let alone external) from health care price signals. They'll cut back on essential care, which is obviously undesirable, as well as elective care. Nor should it be necessary to entertain seriously the notion that people are hyper-rational to the extent that potentially expensive meds in old age will induce them to exercise more or eat fewer fatty foods now.

I've discussed this sort of argument before and dislike it enough that I might as well coin a vulgar term for it: Asshole price theory, or the abuse of Econ 101 supply-and-demand logic to rationalize the unpleasant implications of a policy one just happens to like. In this case, there is no firm foundation for leaping from a desire to transmit a price signal to imposing 100% coinsurance; the only sure consequence is that the rate of coinsurance should not be zero. For that matter, if sending strong price signals were the paramount concern, then it may be a strange choice to insure 75% of costs below the hole, the previous discussion of chronic conditions notwithstanding. Moreover, to the extent that drug expenditures in the hole tend to be incurred for relatively unpredictable acute conditions, those are the sort of expenditures it's traditionally desirable to insure against.

So I'll go out on a limb and suggest that the optimal rate of insurance in the hole is something other than zero. Since that's true regardless of the level of drug prices, at least for any foreseeable level of drug prices, Jeremy is incorrect to say, "Drug prices are the problem, not the doughnut hole." (***)

-------------------

Jeremy then criticizes Krugman for failing to highlight how the benefit kicks back in past the hole.

Anyway, nowhere in his column does Krugman mention that once you get past $5,100, the government picks up 95% of your drug costs. If you are going to say that the program provides a "very poor benefit" for those who need "a lot of help" with their drug costs, it seems dishonest not to mention this...

I'll grant this would have been scrupulously honest, though my counterfactualist riposte is that provision of a benefit for "catastrophic" expenses was the common feature among all of the competing Medicare drug benefit proposals. Since Krugman's counterfactual program would have plugged the hole and more, the conclusion would stand that the inframarginal benefits — the $3,600 out-of-pocket prior to reaching the $5,100 stop-loss level — are poor, if not "very poor," and plainly the result of Medicare beneficiaries being chiseled for the political benefit of the Republicans and for the financial benefit of Big Pharma and health insurers. (That Medicare beneficiaries have been sacrificed in no small part for the benefit of Big Pharma and the insurance industry is a point of agreement among Krugman, Jeremy, and me, among many others.) That the enacted program offers poor bang for the buck is hardly new to regular Krugman readers, for instance, November 21, 2003:

"This is a good bill that will help every Medicare beneficiary," wrote Tom Scully, the Medicare administrator, in a letter to The New York Times defending the prescription drug bill. That's flatly untrue. (Are you surprised?) As the Center on Budget and Policy Priorities points out [PDF, link added], the bill will force millions of beneficiaries to pay more for drugs, thanks to a provision that cuts off supplemental aid from Medicaid. Poorer recipients may find previously affordable drugs moving out of reach.

That's only one of a number of anti-retiree measures tucked away in the bill. It contains several Trojan horse provisions that are clearly intended to undermine Medicare over time — it will allow private insurers to cherry-pick healthy clients in selected cities, and it will heavily subsidize private plans competing with traditional Medicare. Meanwhile, the bill prohibits Medicare from using its bargaining power to cut drug prices; drug company stocks have soared since the bill's details became public.

[...]

And this Medicare bill is very friendly to insurance and drug companies. Senator John Breaux, one of only two Democrats who participated in negotiations over the bill, takes the controversy as a good sign: "No one got everything they wanted." But as Jonathan Cohn points out in The New Republic, drug and insurance companies got exactly what they wanted: no efforts to limit prices, generous subsidies and lots of additional business. For example, insurance companies that offer an alternative to Medicare will not only be able to pick and choose their customers, but will also get 30 percent more per client than the government spends on the average Medicare recipient.

Finally, Jeremy commits a logical foul in implying that the hole is a necessary consequence of the failure to include drug price controls in the legislation:

Krugman knows full well that once the government legislated itself out of any capacity to control or negotiate prices, something had to be done to keep the response of drug companies (who, granted, helped a lot in writing the bill) from simply multiplying the price of their drugs in response to the benefit.

Again, imposing discipline on drug companies via patents' cost sensitivity doesn't require 100% coinsurance, and with three coinsurance rates to vary, among other parameters of the benefit design, there are degrees of freedom that would allow the hole to be avoided for just about any realistic level of program cost. In any event, Krugman has pointed out briefly in Friday's column and previously at greater length (as cited above) that the shortcomings of the benefits ultimately stem from the messed-up political imperatives that drove the legislation.

Overall, I find Krugman to be both partisan and correct.

-----------------

(*) I actually prefer to look at it as one benefit with a variable coinsurance rate.

(**) Typical lifetime benefit caps in private health insurance contracts are in the seven figures, though caps for certain specific services, e.g. outpatient mental health care, may bind much more commonly. Insofar as reaching a lifetime benefit cap is rare but conditionally devastating, not to mention expensive to privately insure, public insurance (or at least reinsurance) for truly catastrophic expenditures is a good idea.

(***) Otherwise, it's an "I want low prescription drug prices, and a pony!" argument.