Currently the vast majority of tax incentives operate through deductions or exclusions, which link the size of the tax preference to a household’s marginal tax bracket. Higher-income taxpayers, who are in higher marginal tax brackets, thus receive larger incentives than lower-income taxpayers. This Article argues that providing a larger incentive to higher-income households is economically inefficient unless policymakers have specific knowledge that such households are more responsive to the incentive or that their engaging in the behavior generates larger social benefits. Absent such empirical evidence, all households should face the same set of incentives.
Discussing our Hamilton Project paper, Orszag brought this point up with respect to the employer-sponsored health insurance tax exclusion. It’s hard to make an efficiency argument for the regressivity of that exclusion, and it’s tempting to raise its inefficiency as justification for the Cadillac tax …
In our Hamilton Project paper, Nicholas, Amitabh, and I propose a reference pricing scheme as one part of a package of ideas to manage the rate and composition of innovation of health care technology. If you’re already thinking about CalPERS’ reference pricing approach—about which I’ve written a lot (here’s one post)—you’re on the wrong track. Though we’ve never seen the distinction clearly drawn in the literature, there are two notions of reference pricing out there, and we mean the other one.
That other one has been used for prescription drugs in Canada (British Columbia), Germany, Norway, Spain, and other countries, as well as two employers (in the US and Canada)—see the systematic review of reference pricing by Lee et al. It also arises in the clever proposal by Pearson and Bach.
The basic idea is to group therapies (like drugs, but also procedures or technologies) according to their therapeutic effect. Then, one sets a one price—the reference price—for each group in some way. It could be the price associated with the most cost-effective treatment, for instance. Notice, however, that that price could vary across providers, maybe paying a higher quality provider more, another less.
I call this vertical reference pricing. It’s set within provider, but applies across technologies grouped by therapeutic effect. That’s very different from what I call the horizontal reference pricing of CalPERS, which sets one price for, say, a knee replacement and applies that across all providers.
Let’s play out the differences between vertical and horizontal reference pricing for the three prostate cancer treatments offered at three different hospitals and at prices shown in the chart below. (Yes, there are other prostate cancer treatments, like prostatectomy and watchful waiting, but for simplicity let’s pretend for the moment that these are the only three we know of.)
Horizontal (CalPERS style) reference pricing would pick one price for prostate cancer treatment, perhaps something like the “average bundle” price of $13k at hospital C, which hospital C has agreed to accept for payment in full for any prostate cancer treatment (no out-of-pocket cost for the patient). It would then offer that price to any hospital. If a patient received prostate cancer treatment at hospital A or B, he might pay more out of pocket for it because some therapies offered at those hospitals cost more than $13k.
This puts considerable pressure on the patient to shop for a hospital that will accept the reference price as payment in full or for the patient to weigh the quality-price trade-off at hospitals that won’t. It puts considerable pressure on the insurer to set a price low enough to reap savings but high enough such that there are sufficient high-quality and affordable choices in the market, something not all markets can support. Adhering to a reasonable access standard would also impose an ongoing, management overhead cost. These are among the critiques one can level at horizontal reference pricing (see Frankford and Rosenbaum).
Many of those critiques do not apply to vertical reference pricing, however. Let’s play out the prostate cancer example to see why. For example, imagine pegging the hospital-specific vertical reference price to the most cost-effective prostate cancer treatment (among the three in the chart). Let’s say that’s intensity modulated radiation therapy (IMRT).* Vertical reference pricing would pay the hospital-specific IMRT price ($20k, $15k, $10k at hospitals A, B, C, respectively). If a patient wanted proton beam therapy at any of these hospitals, he’d pay the (hospital-specific) difference.
The advantage of this kind of vertical reference pricing is that it guides the patient toward more cost-effective technologies, and fully covers it or anything cheaper at any provider (or any provider in the insurer’s standard network). One need not shop around if one is OK with cost-effectiveness. It also guides manufacturers toward development of cost-effective technologies.
There are, of course, many variants, like pegging the vertical reference price to the most demonstrably effective therapy (ignoring cost). This at least spares the insurer the cost of more expensive and unproven therapies, but partially compensates any patient that wants it. Our proposal is closer to this approach, but we bring in the idea that no price (per QALY) should be greater than a cost-effectiveness threshold.
We also propose that Medicare pay whatever it otherwise would for cost-effective therapies priced below the reference price. An implication of this approach is that something as cheap to Medicare as, say, brachytherapy for prostate cancer remains just that cheap. It is not paid for at the reference price. Under horizontal reference pricing, on the other hand, it would get reimbursed at the same reference price as any other prostate cancer treatment, which would overcompensate providers for it.
I refer you to our paper for the details on reference pricing and our other proposals, which complement it. You can also attend the event at Brookings tomorrow at which it will be discussed.
* I don’t know if that is, in fact, the most cost-effective therapy. Let’s just pretend it is for this example.
On Wednesday, Nicholas Bagley, Amitabh Chandra, and I are presenting a paper at Brookings as part of its Hamilton Project. Our paper explores ways we might change the market signals—shaped by our laws, regulations, and insurance designs—we send to manufacturers about what health care technology to develop, pushing them more strongly toward cost-effective therapies.
There are several reasons why you might want to read the paper. In this and forthcoming posts, Nicholas and I will convey them. In this first post, I want to encourage you to read it because it elaborates on limitations of an idea Amitabh and I enthusiastically wrote about on The Upshot:
Health plans could define themselves at least in part by the value of technologies they cover, an idea proposed by Professor Russell Korobkin of the U.C.L.A. School of Law. For example, a bronze plan could cover hospitalizations and visits to doctors for emergencies and accidents; genetic diseases; and prescription drugs that keep people out of hospitals. A silver plan could cover what bronze plans do but also include treatments a large majority of physicians find useful. A gold plan could be more inclusive still, adding coverage, for instance, for every cancer therapy shown to improve patient outcomes (no matter the cost) as long as it was delivered at a leading cancer center. Finally, a platinum plan could cover experimental and unproven cancer therapies, including, for example, that proton beam.
This way, nothing would be concealed or withheld from consumers. Someone who wanted proton-beam cancer treatment coverage could have it by selecting a platinum policy and paying its higher premiums. Someone who did not want to pay higher premiums for lower-value care, in turn, could choose a bronze or silver plan.
In our piece, we noted the adverse selection problem this would invite.
[A]s people become sick, they will prefer plans that cover more treatments, including experimental ones. As sick people disproportionately choose more generous plans, their expenses and premiums will have to rise. This phenomenon, known as adverse selection, is familiar in most health insurance markets, including those for employer-sponsored plans, private plans that participate in Medicare and in the Affordable Care Act’s new marketplaces. One common way to address it is to permit individuals to switch plans only once per year, during an open enrollment period. This locks people into their choice for some time, so they can’t suddenly upgrade their plan after getting sick. If a once-per-year enrollment period proves insufficient in this case, a longer period could be imposed.
We now understand that the adverse selection problem would be even more severe than we had imagined. In our Hamilton Project paper we explain why.
Consider  a young, married couple with no children [and] with a modest demand for technology, both because they’re healthy and because they value exotic vacations more than exotic treatments. They select the low-technology (high cost-effectiveness) option and use the savings to travel abroad. Now suppose that they have a child who needs treatment for cystic fibrosis. Novel therapies for this condition have an incremental cost-effectiveness ratio in the hundreds of thousands of dollars (Whiting et al 2014). The family may rationally want to switch from their plan with stingy coverage rules to an expansive plan that covers high-cost therapies with low cost-effectiveness. Since the reason for the parents’ plan switch is to offset the cost of a particular therapy, the plan to which they switch will incur its cost with certainty. Because the parents’ change in technology preference is intimately linked to a change in diagnosis, applying standard, diagnosis-based risk adjustment approaches would spread this additional cost to low-technology plans. Forcing low-technology plans to pay for the expensive technology they exclude would destroy the entire point of this kind of market.
There are solutions to this, but they’re grossly unpalatable. They include a return to pre-existing condition exclusions or very long, possibly lifetime, commitments to level of insurance for health care technology. Thus, we’re skeptical that a market centered on plan competition on the cost-effectiveness of health care technologies is culturally or politically feasible. But, we have other ideas! Read the paper for more.
If you’re a fan of any of the folks in bold below, or the topics they’ll discuss, and are in the DC area, you might want to register for Wednesday’s Brookings’ Hamilton Project event.
On October 7, The Hamilton Project will host a policy forum addressing  economic challenges in an evolving health-care market, with a focus on three new papers released in conjunction with this event. Opening remarks will be delivered by former U.S. Treasury Secretary Robert E. Rubin. Jason Furman, chairman of The Council of Economic Advisers, will deliver framing remarks. Event participants will include: Peter Orszag, nonresident senior fellow in Economic Studies at Brookings and vice chairman of corporate and investment banking at Citigroup, Inc.; Rick Pollack, president and CEO of the American Hospital Association; Niall Brennan, chief data officer, the Centers for Medicare & Medicaid Services; Julie Rovner, Robin Toner distinguished fellow and senior correspondent, Kaiser Health News; Martin Gaynor, E.J. Barone professor of economics and health, Carnegie Mellon University; and Dan Durham, executive vice president, America’s Health Insurance Plans.
The first panel will discuss a new paper by Professors David Dranove, Craig Garthwaite and Christopher Ody (each of Northwestern University’s Kellogg School of Management) which proposes a tradable credit system of community benefits in the non-profit hospital sector. The second panel will focus on two new papers: “Health Insurance Markets and Medical Technology Coverage,” by Professors Nicholas Bagley (University of Michigan), Amitabh Chandra (Harvard University), and Austin Frakt (Boston University) and “Improving Decision Making in Health Insurance Markets” by Professors Benjamin Handel and Jonathan Kolstad (both of University of California, Berkeley).
All papers address challenging problems. In particular, ours includes proposals to reform the Cadillac tax, as well as drug and other health care technology pricing and coverage. (These are kinda hot issues, so …)
[S]everal studies have suggested that increased choice may not be beneficial to decision makers. Despite the greater likelihood of a better option being available, a larger number of choices may lead to choice overload, greater regret, and more indecision. This has led some to suggest that choice sets should be restricted. From a practical standpoint, all proposals calling for restricting a choice set face the criticism of being paternalistic in determining how choices are restricted.
Instead of attempting to restrict the choice set, we seek to identify whether restructuring choice architectures can enhance decision quality while still maintaining the size of the choice set. […]
Our findings essentially push the paternalistic discussion associated with choice overload back one level. Our work suggests that more, but not all, people would select better options with a sequential tournament; however, this choice architecture may be the least preferred of those we consider. Therefore, in some cases, policymakers or others designing a choice problem may wish to impose an unpopular procedure in order to improve decision-making quality.
It’s a fascinating paper with impressively careful methods.
In my third post in my series on how bad people are at choosing health plans, I continue to summarize studies relating to commercial market plans, which I started in my second post. (See post 1 for research pertaining to Medicare plans.) My third post is on the AcademyHealth blog. Go read it!
I’ve been standing at my workstation for several years, and there’s no way I’d go back to sitting all day. I’m much more comfortable standing (less back/neck/arm discomfort, resulting in a better mood). But that’s as close as I’d get to a health claim. I don’t think it’s making me fitter or adding years to my life. If you’re looking for a massive productivity or health boost from your standing desk, a 2014 systematic review may disappoint you.
It concludes that standing and treadmill desks probably offer some health value, more so for obese users, but the evidence isn’t strong and there are hedge words all over the conclusion (my emphasis):
Based on the empirical evidence of current literature, this review concludes that standing and treadmill desks are potentially useful in reducing workplace sedentariness while having a positive influence on workplace stress and overall mood. The treadmill desk provides the greatest physiological improvements and is most beneficial for overweight and obese participants. However, the use of a treadmill desk results in larger decreases in work productivity and motor abilities than the standing desk.
Standing desk use does not elicit the same physiological impact as the treadmill desk but does result in the least change in productivity and motor abilities. Of the standing desks, a sit–stand desk seems to provide the most benefit allowing the employee to adjust their desks throughout the day. The standing-only desk could potentially result in additional complications with musculoskeletal conditions and feelings of fatigue and discomfort.
Overall, current evidence suggests that both standing and treadmill desks may be effective in improving overall health considering both physiological and mental health components. However, at present there still exist substantial gaps in the research to fully comprehend the utility of each type of desk to promote health.
I thank Aaron for sending me this review. I still love my standing desk.
I’m sure few TIE readers have failed to notice that there’s a ton of discussion of drug prices and price control policies these days. Could the mere consideration of such policies cause price reductions, perhaps by manufacturers trying to avoid the bad PR that could turn discussion into legislative action?
Work by Ellison and Wolfram suggests the answer is “yes.” They examined drug prices during the run up to and through the early 1990s efforts at national health reform. One of the policy ideas in 1993 was to push drug prices downward via purchasers with massive market clout. In particular, Medicare was to add prescription drug coverage to its benefit package and use its massive size to negotiate discounts. (Sound familiar?)
The data suggest that this spooked the drug industry. While the health reform package was being formulated, pharma stocks fell 40%, Ellison and Wolfram wrote. Here’s their look at what happened to prices, by various measures (defined under the chart):
In the chart, “PPI” and “CPI” are the producer and consumer price index for pharmaceutical products; “CPI overall” is the CPI for all products; “Top 106” is a set of the 106 largest revenue drugs in the 1990s; “antibiotics” are “virtually all” antibiotics sold from 1990-1996.
From 1991 through 1996, overall CPI growth was steady, but by all drug measures, drug price growth fell, and dramatically so just as health reform was being formulated. In particular, broad measures of drug price growth were nearly pegged to overall inflation growth from 1993-1996.
Growth in drug R&D took a hit too, as shown in the following chart. As a percent of sales, it flattened in the early 1990s. Annual change went negative in 1994.
The authors conclude,
The results presented here suggest that there was a political component to pharmaceutical pricing during the health care reform debates. In particular, the firms we identified as more politically sensitive were more likely to engage in coordinating pricing, consistent with a pledge many firms made not to raise prices more than the rate of inflation.
A caveat is that most of the action was restricted to wholesale prices, though some filtered through to transaction prices. A lesson is that merely the threat of policy change can affect prices and R&D investment. Talk is still cheap, but perhaps not as cheap as we may think.
Austin and Aaron are participants in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for sites to earn advertising fees by advertising and linking to amazon.com.