- Government has incentives to focus far more heavily upon budget outlays than patient wellbeing in the operation of its health programs.
- So as to obtain medical devices and equipment for the beneficiaries of Medicare and Medicaid, the federal government uses auctions that are poorly designed, resulting in prices lower than would be observed in a competitive market.
- These artificially-low prices reduce private-sector incentives for investment in the research and development of new and improved medical technologies.
- The technical literature suggests that this decline in research and development investment will be about 12 to 15 percent between 2011 and 2020.
- Under very conservative assumptions, the resulting adverse effect upon patient wellbeing would be a decline in expected life-years for the U.S. population of about 500,000 annually, concentrated upon particular population subgroups.
This decline would be equivalent to an economic loss of about $50 billion per year, a figure substantially greater than the entire U.S. market for medical devices and equipment.
Reform of the auction process should be a high priority for policymakers.
Introduction. Government does not have patients. It has interest groups, a crucial distinction that yields important implications for the design and operation of such government benefit programs as Medicare and Medicaid. In particular: Dollars not spent on a given constituency can be spent on others, yielding powerful incentives to reduce outlays on any given program so as to expand the number of programs. Because government does not have patients, it is costs that are measurable—budget outlays—as opposed to costs that are hidden that can be predicted to drive decisionmaking about program design, benefits, and the choices among alternative goods and services. An example of such a hidden cost is the higher marginal value (if any) of a more-expensive alternative (e.g., a costlier drug), as perceived by the participants in the government programs, but perhaps perceived only weakly if at all by government decisionmakers, who have strong incentives to place greater weight on budget savings and less weight on greater medical effectiveness. That is an example of a short-run hidden cost. An example of a long-run hidden cost: research and development investment in new and improved medical technologies that is not made because of prices expected to be lower than economically efficient, an outcome to which we return below.
Because government health programsagain, Medicare and Medicaid in particularuse auctions for the acquisition of medical equipment and devices, this incentive on the part of government agencies is likely to shape the design of the auctions; and the choice among alternative designs can affect prices and supplies in important ways. Accordingly, government as a buyer of goods and services through auction mechanisms has incentives to design auctions in ways that favor lower-priced goods and services over higher-priced ones, and that can yield prices lower than competitive. However counterintuitive it may seem, prices can be inefficiently low as well as inefficiently high, outcomes that can have serious adverse effects. In the context of auctions for the procurement of medical equipment and devices, this governmental bias is likely to yield the delivery of goods and services lower in quantity and/or quality (or effectiveness) than otherwise would be the case, and a reduction in incentives for investment in research and development.
The Centers for Medicare and Medicaid Services (CMS) established in late 2009 an auction system for medical devices and equipment; at the beginning a pilot program, the system will be expanded to one hundred cities in 2012. It is clear from the technical literature that this auction design is seriously flawed, in ways that can be predicted to yield prices substantially lower than those that would be observed in a competitive market.1 Because incentives to invest in the research and development of new medical technologies are driven by expected prices and economic returns, this downward bias in auction prices is virtually certain to reduce such investment, with respect to which the system in effect acts as an implicit tax.
Effects on Research and Development Investment. The question to be addressed is: How important prospectively is that adverse investment effect in terms of research and development investment, the flow of new and improved medical technologies, and the wellbeing of patients? Data from CMS show that in 2009 total national expenditures on durable medical equipment was about $34.9 billion, of which about $10.4 billion, or about 29.9 percent, was paid by federal non-defense programs. CMS projects that the federal non-defense market share for durable medical equipment will rise from about 32.3 percent in 2011 to about 39.0 percent in 2019.2
The economic literature on the effects of the CMS auction design yields a range of predicted effects in terms of price suppression. The lowest of those findings is 33.4 percent.3 This allows us to compute an implied price effect per year as a function of the projected federal market share.4 Table 1 presents those projections for the period 2011 through 2019.
A simple econometric model of research and development investment in medical devices and equipment can be used to estimate these price effects upon research and development investment.5 Table 2 presents those estimated effects for 2011-2020.6
The aggregate projected adverse investment effect is about $2.1 billion in 2011, rising to about $3.1 billion in 2020, or about 12 to 15 percent. Investment growth between 2010 and 2020 falls from 2.24 percent annually to 0.8 percent annually.
Effects on Patient Wellbeing. The scholarly literature on the economic returns to investment in medical technologies suggests that research and development investment in new and improved pharmaceuticals yields, roughly, an increase of an expected life-year for each investment of about $2000 (in 2010 dollars).7 There does not appear to be available in the literature a similar analysis examining the impact of such investment in medical devices and equipment. But the value of investment in medical technologies is driven ultimately by the perceived value of increased longevity and health. Medical technologiespharmaceuticals, devices, equipment, physician training, etc.can be complements or substitutes, or, often, both, and the medical marketplace (in the simple case) has incentives to find the most cost-efficient mix of technologies to achieve a given medical outcome.8 This means that the benefits of investments in different medical technologies should be approximately the same over time, so that it is reasonable to use the Lichtenberg findings to derive rough estimates for devices and equipment.
If we assume from Table 2 that the CMS auction system reduces investment by (very conservatively) $1 billion per year, the loss in expected life years would be about 500,000 annually. (Note that this adverse effect would not be spread uniformly across the population.) If we assume $100,000 to be the value of an expected life-year, the economic cost of the CMS auction system is about $50 billion per year. 9 That figure is substantially greater than the entire U.S. market for medical devices and equipment. 10
Conclusions. It is clear from the technical literature that the CMS auction design for medical devices and equipment is seriously flawed, in ways that can be predicted to yield prices substantially lower than those that would be observed in a competitive market. Those prices are inefficient, as they do not provide signals or incentives to produce goods and services optimal in number, quality, or characteristics.
The experimental literature on the CMS auction design is unambiguous: It yields prices too low. This outcome results from the design features: Bids are not treated as binding commitments, the contract price is the median among the winning bids rather than the bid reflecting marginal cost, the composite bid system of averaging over heterogeneous products skews bids in ways driven by perceived errors in demand projections, and the allocation of market shares is opaque. These problems yield prices about two-thirds to one-third below the competitive price.
Because incentives to invest in the research and development of new medical technologies are driven by perceived returns, this downward bias in auction prices is virtually certain to reduce such investment, with respect to which the system in effect acts as an implicit tax. Our finding is that such investment will be reduced by about 12 percent to 15 percent annually, or about $2.1 billion to $3.1 billion annually during 2011 through 2020. Annual investment growth would fall from about 2.24 percent to about 0.8 percent. By analogy with the estimates available in the literature for pharmaceutical investment, this investment loss would cause, conservatively, a loss of about 500,000 expected life-years each year, the economic cost of which would be about $50 billion per year, a figure substantially greater than the entire U.S. market for medical devices and equipment. The sheer magnitude of this adverse economic effectderived from very conservative assumptionssuggests strongly that reform of the CMS auction process should be a high priority for policymakers.
1. The CMS auction design yields prices lower than competitive, an outcome that results from the design features: Bids are not treated as binding commitments, the contract price is the median among the winning bids rather than the bid reflecting marginal cost, the composite bid system of averaging over heterogeneous products skews bids in ways driven by perceived errors in demand projections, and the allocation of market shares is opaque. These problems yield prices about two-thirds to one-third below the competitive price.
2. Source: CMS at https://www.cms.gov/NationalHealthExpendData/03_NationalHealthAccountsProjected.asp#TopOfPage; and data adjustments by the author to make the projections consistent with the CMS data for 2009.
3. See Brian Merlob, Charles R. Plott, and Yuanzun Zhang, The CMS Auction: Experimental Studies of a Median-Bid Procurement Auction With Non-Binding Bids, California Institute of Technology Social Science Working Paper 1346, April 2011, at https://www.hss.caltech.edu/SSPapers/sswp1346.pdf.
4. These projected price effects are conservative because they assume away the likely effect of federal prices on non-federal prices.
5. The details of this model are presented in Benjamin Zycher, Medicare Auctions for Durable Medical Equipment: Price Suppression and Research and Development Investment, Pacific Research Institute monograph, June 2011, available at https://www.pacificresearch.org/docLib/20110614_MedDevices.pdf, at 18-20.
6. The annual average compound growth rate in research and development investment for medical devices and equipment for 1999-2007 was about 2.24 percent.
7. See Frank R. Lichtenberg, Sources of U.S. Longevity Increase, 1960-2001, Quarterly Review of Economics and Finance, Vol. 44, No. 3 (July 2004), pp. 369-389. A somewhat different analysis is presented by Kevin M. Murphy and Robert H. Topel, The Economic Value of Medical Research, in Kevin M. Murphy and Robert H. Topel, eds., Measuring the Gains from Medical Research: An Economic Approach, Chicago: University of Chicago Press, 2003, pp. 41-73. They estimate a net return of about $1.6 trillion annually from an annual research and development investment of $35 billion, a rate of return of almost 4500 percent.
8. In more technical language, the market has incentives to equate the marginal returns to alternative investments. Accordingly, as a first approximation, it is reasonable to assume that such returns—that is, the medical benefits that investments in different medical technologies yield for patients—tend to converge. This means that the Lichtenberg finding for pharmaceutical investment is appropriate, again as a first approximation, for other technologies as well.
9. Murphy and Topel, op. cit., fn. 7, estimate the value of an expected life-year at $160,000.
10. See data at Centers for Medicare and Medicaid Services, at https://www.cms.gov/nationalhealthexpenddata/downloads/tables.pdf, Table 11.