Did Part D Work?

Mark Duggan and Fiona Scott Morton published a paper at NBER with this general conclusion:

Using data on product-specific prices and quantities sold in each year in the U.S., our findings indicate that Part D substantially lowered the average price and increased the total utilization of prescription drugs by Medicare recipients. Our results further suggest that the magnitude of these average effects varies across drugs as predicted by economic theory.

Even though they were only using one year's worth of data—or perhaps because of it—they concluded that the program has been a success for the most common drugs. I posted this list at MU, but it bears repeating here:

Lipitor, Zocor, Prevacid, Nexium, Zoloft, Epogen, Celebrex, Zyprexa, Neurontin, Procrit, Effexor, Advair, Paxil, Norvasc, Pravachol, Plavix, Allegra, Wellbutrin, Oxycontin, Fosamax, Vioxx, Singulair, Protonix, Actos, Ortho, Aciphex

Duggan and Scott Morton also found that—for the "small subset of 'protected' therapeutic classes" that all providers were required to carry—the Part D prices to consumers actually rose. The authors explain this as a standard of economic theory:

According to Part D regulations, there are six “protected” therapeutic classes in which PDPs must be less aggressive with their formularies than in other therapeutic areas. All products in the HIV, anti-cancer, anticonvulsant, immunosuppressant, antipsychotic, and antidepressant categories must be included in all Part D formularies. While a PDP cannot exclude any drug in these categories, it can create financial incentives or administrative hurdles to affect a patient’s choice of drug....We do not know whether the restrictions applied to these classes have a measurable impact on the behavior of PDPs because in the first year of the program it was not clear how much CMS would oversee formularies. If restrictions are binding, their effect will be to reduce Part D’s effect on the substitutability among drugs (lower [gamma-sub-g]) and therefore reduce the PDP’s ability to extract manufacturer discounts.

English version: without being able to threaten to exclude a drug from coverage, and not being certain about whether they would be permitted to classify a drug as more expensive than a counterpart under their specific plan, the drug companies could not bargain effectively with drug manufacturers.

Which makes perfect sense, until you consider that the price to Medicare recipients of those drugs went up.

Imagine the negotiations. "You charge $1,000 for that drug." "Yes, but for you, $1,005." "Done."

Duggan and Scott Morton do present some caveats

To the extent that plans become more or less successful at negotiating prices in future years, the results may of course change. Secondly, we are unable to measure any ex post rebates which PDPs may have been able to negotiate and which affect net prices to PDPs. Such rebates do not appear on the invoice, which is the source of IMS data, and might be causing prices to be even lower than those measured here. If rebates are present, our estimates are a lower bound to the price reductions achieved by PDPs.

Translation: Some PDPs may be making more money than we think at the current levels.

The other rebates that we do not measure are Medicaid rebates paid by manufacturers to the Medicaid program. Dual eligibles’ pharmaceutical purchases under Medicaid automatically generated this rebate. Once dual eligibles move into Medicare Part D plans, their pharmaceutical purchases occur at different prices, which is what we document here, but they no longer trigger automatic rebates. Any study of the total cost of Part D to the government would want to consider both sets of rebates.

Translation: While the PDPs may have earned more, the government may have spent more (rebates not received).

The last two possible results would be similar to those expected by many economists who looked at the form of Part D, where the largest buyer (the Government) was prevented from using its buying power, but obligated to foot the bill for private companies that, individually and probably even collectively, would not be able to negotiate with the same influence.

More worrisome than that this conclusion should be expected is what might be expected to happen if the PDPs were rational. Again, this would come from standard economic theory, though it is not discussed explicitly by Duggan and Scott Morton.

To be direct about it, the PDPs in Medicare Part D each has a steep learning curve, effectively creating a switching cost for the consumer. That, in turn, will enable each PDP to retain its consumer base, even while increasing the prices it charges.

Health Economists especially are fond of talking about the "full cost" of something. If it would take me twenty or thirty hours to select a replacement PDP, the that "cost" will keep me from switching, even if I end up paying a few dollars more for a prescriptions.

Contrast this with, say, automobile insurance. The terms are all similar, and I can spend "15 minutes" getting a quote from GEICO (or three or four from Progressive) that I know is essentially the same coverage as my current provider.

I may not know how well the insurer will respond to me, and I may not know if they can provide the other policies I need (home, life, etc.), so there may be minor externalities (e.g., having to write different checks at different times to different insurers for different policies). But there will be nothing on the order of the switching costs currently associated with Medicare Part D.

Duggan and Scott Morton have done a service in indicating that Part D has gotten more people using more drugs.* And they have so far shown that economic theory appears to be holding in a real-life situation.

If economic theory continues to hold, we should expect that profit margins will grow over time, in reaction to the high switching costs that are built into the program.

We can hope that will not be so, but Mark Duggan and Fiona Scott Morton have not indicated that would be the way to bet.

*This is also the lesson of the Massachusetts universal health plan, but that's for another post, though I note that the differing reactions to the two situations from some of the think tanks is interesting in itself.

Original post appeared on the AngryBear blog at http://angrybear.blogspot.com/2008/05/did-part-d-work.html