Improved market access

One possible advantage of price discrimination is that higher income individuals (or groups) can subsidize prices for less affluent individuals (or groups), hence improving access for the latter. If a drug company can use price discrimination to open up new markets and thus increase market output, then society as a whole may be better off (Schmalensee, Reference Schmalensee1981).Footnote 3 This argument has been used in the context of the international market for pharmaceuticals where drug companies can price discriminate between industrialised and low-income countries, for example, for HIV/AIDS drugs. This allows low-income countries access to drugs they would most likely not have been able to afford under a uniform global price.

While this is possibly a valid argument in the international setting, it is, unclear how price discrimination across the various federal agencies results in expanded market access for some federal beneficiaries. Transferring the ‘improved market access’ argument to the US government setting would suggest that the VA and DoD, constitute the most price sensitive groups, the equivalent of the ‘weak’ markets in low income countries. Similarly, Medicare and Medicaid, currently paying more, are in the position of the ‘stronger’ markets.

While it may be that access to drugs would be even further restricted for poorer countries if higher income countries were to insist on a uniform global price, it is unlikely that VA and DoD would be unable to afford necessary drugs if Medicare and Medicaid were to pay lower prices. Furthermore, there is currently no transfer mechanism for higher prices paid to drug companies by Medicaid and Medicare to translate into lower prices for the VA or DoD. Indeed, price increases following the introduction of Part D in 2006 led to higher company revenues and not lower prices for the VA and DoD markets (Frank and Newhouse, Reference Frank and Newhouse2007).

Economies of scale and learning

Price discrimination can positively affect welfare if total output increases. This is in comparison with a situation of no price discrimination. One way that this could happen is if there are economies of scale for the good or service in question. This is because unit costs decrease as a function of accumulated experience or reduced incremental production costs (Hausman and MacKie-Mason, Reference Hausman and MacKie-Mason1988). Once a pharmaceutical product has been developed, scale and learning economies are unlikely to result in lowering average marginal costs as output increases thanks to discrimination opening up new markets. In this sense, although economies of scale do exist in the research and development components of pharmaceutical research – for example, in navigating complex clinical trial requirements for large firms, or by tapping into larger companies’ expertise through licensing agreements for smaller companies – once a product has been launched, such economies of scale do not (at least not directly) affect the, already very low, marginal cost of production. For example, factors such as regulatory barriers, patent issues and the cost of setting up and maintaining distribution networks (in addition to the threat of arbitrage) make poorer countries less attractive markets even when rich ones pick up the cost of Research and Development (R&D) through much higher than marginal cost prices.

Enhanced competition

Economic analyses of price discrimination traditionally focus on industries operating in a monopoly setting. However, the market for branded pharmaceuticals is closer to an oligopoly because single-source branded drugs often face competition from therapeutic substitutes or other treatment modalities. In this setting, the effects of price discrimination on total welfare, according to economic theory, are more sensitive to assumptions about the shape of the demand curves of the different market segments. In certain cases, price discrimination can lead to welfare gain through intensified competition.

For this to occur, competing firms must differ in their assessment of a particular government agency’s price sensitivity for the same type of drug. As an example, the manufacturer of a branded anti-hypertensive drug might raise its VA price while the manufacturer of a competing anti-hypertensive drug might reduce its VA price, based on differing assessments of the government’s price sensitivity. This ‘best response asymmetry’ is usually driven by consumers’ strong brand preferences (rather than price). According to economic theory, different assessments of brand loyalty would trigger competition between the therapeutic substitutes and lead to price reductions across the board (Holmes, Reference Holmes1989; Corts, Reference Corts1998; Armstrong, Reference Armstrong2006). However, in our case, federal (or state) agencies are the consumers and they are forbidden by law from exercising brand loyalty, making this scenario less applicable in the government context.

Better incentives for R&D

Price discrimination can support future R&D investment if it allows for higher aggregate spending on drugs by the federal government (Hausman and MacKie-Mason, Reference Hausman and MacKie-Mason1988; Danzon, Reference Danzon1997). We return to the ‘R&D incentives’ argument towards the end of this study.

For price discrimination to make economic sense, observed price differentials need to reflect underlying demand elasticities (first condition). Does this theoretical framework, developed for individual consumers making choices based on their willingness to pay, also apply in the government context? A relevant question is whose demand curves should be considered when drug companies are selling to government programmes – beneficiaries or government agencies?

To the extent that price sensitivity is an inverse function of income, government agencies that benefit the poor should have higher elasticities of demand. However, in the context of the health care market, demand elasticities are distorted by public and private insurance, imperfect agency on behalf of the physician and information asymmetry, and further compounded by the high cost to patients (or clinicians) of accessing information on the relative value of treatments. Co-pays are relatively small, particularly in the case of Medicaid, and premiums rare. For example, given the initial deductible in most plans, and an up to 100% coinsurance if a drug becomes excluded from a formulary, Medicare beneficiaries are likely to have higher elasticities of demand than Medicaid ones.Footnote 4 In most cases, the federal beneficiaries are (at least in part) insulated from the price differentials paid by the various government agencies.

The first condition of price discrimination could still apply if individuals chose to use certain government agencies based on their own sensitivity to price and value. Government agencies utilise different measures to control prices and utilisation such as formularies, preferred drugs’ lists, prescription guidelines and mandated rebates. To the extent that some beneficiaries can choose between various government agencies to receive their drugs, the choice may reflect individual preferences. In these special cases, price discrimination would be justified by underlying differences in price sensitivity of beneficiaries. However, most beneficiaries are not eligible for drug benefits under multiple federal programmes.

In conclusion, neither volume-based discounts nor price discrimination are good rationales for the observed drug price differentials. It seems that legislative and organisational structures that evolved differently over the years, led to distinct purchasing approaches, providing the government with very different types of bargaining power and creating different incentives for the industry. These institutional arrangements dominate economics, distorting the government market for pharmaceuticals with important economic and ethical implications.

Structural and organisational obstacles

Given their weak grounding on economic theory or ethics, drug price differentials across government purchasers are more the products of historical accidents and subsequent political compromises. This evolution has in turn generated and is currently sustaining distinct political constituencies and lobbying parties. There are now significant bureaucracies built around each programme, with different Congressional committees carrying responsibility for VA, DoD, community health centres, Medicare or Medicaid.

Change is thus likely to be opposed not only by drug manufacturers concerned about the ability of the government to negotiate lower prices and PBMs administering Part D plans,Footnote 9 but also government agencies worried about the implications of a rise in spending in their own programme, even if overall government spending remains unchanged. Consolidation may result in some of the bureaucratic and administrative structures becoming redundant, another ground for opposing reform. Perhaps one of the greatest barriers is the myriad of Congressional committees that could lose jurisdiction over drug spending. These political and bureaucratic constraints may be the primary reason for the continuation of differential drug prices.

Aversion to administered prices

There is a general concern that if government determines drug prices, this could create price distortions with possible adverse implications for the health of drug companies. But, with appropriate access to information on comparative effectiveness of alternative technologies, it is possible for the government to increase the value of billions of taxpayers’ expenditures on pharmaceuticals. However, this is subject to the institutional arrangements for prioritising, generating and using comparative effectiveness research. Access to information is necessary (but not sufficient) for government to make informed purchasing decisions, especially in the current climate where decision makers are explicitly prohibited (by statute) from using this information to make coverage decisions (Chalkidou et al., Reference Chalkidou, Tunis, Lopert, Rochaix, Sawicki, Nasser and Xerri2009).

The threat of cost shifting or price distortions in the private sector

Cost shifting exists when a firm raises its prices to one buyer because it lowers the price to another buyer (Morrisey December, Reference Morrisey1996). Two conditions must hold: (a) the firm has enough market power to cost shift (i.e. there is little or no competition) and (b) the firm has not been fully exercising this power (Morrisey, Reference Morrisey2003). Patent protection suggests that the first condition is met. Sole source drugs with no therapeutic substitutes have considerable market power to set the price without significant market constraints (Frank and Newhouse, Reference Frank and Newhouse2008). For the second condition to apply, drug companies must be currently offering some government agencies lower prices than they could obtain if they were fully to use their market power. In other words, the drug companies are currently not charging the full amount they could to certain government agencies.

While higher prices to private insurers have been causally attributed to provider rate reductions by Medicare and Medicaid, the extent to which such effects of administered prices on private sector rates fit the conventional economic definition of cost-shifting has been questioned (Ginsburg, Reference Ginsburg2003). Irrespective of whether cost shifting goes against a firm’s self interest, what matters to policy makers is whether such behaviours have been observed in the healthcare setting, under what circumstances and what their financial impact has been. We examine the empirical evidence of cost shifting or distortionary effects caused by administered prices, in the ‘natural experiment’ that was the introduction of the most favoured customer clause for Medicaid reimbursement of pharmaceuticals (Omnibus Budget Reconciliation Act of 1990 OBRA-90). It has been argued that this policy was directly responsible for higher drug prices for private purchasers, a conclusion often cited as a key reason why government should avoid, in the future, using its bargaining power to get lower prices (Scott-Morton, Reference Scott-Morton1997; Kyle and Ridley, Reference Kyle and Ridley2007; Hogberg, Reference Hogberg2007).

In the early 90s the General Accounting Office carried out two analyses, neither of which supported the cost-shifting hypothesis (Government Accountability Office, 1991, 1993).Footnote 10 In 1996, a CBO report attributed the reduction in the number of large price discounts to private purchasers observed after 1991 to Omnibus Budget Reconciliation Act (OBRA’s) inclusion of a provision that guaranteed Medicaid the ‘best price’ obtained in the private sector. However, without data from the period that preceded the legislation, no causal relationship could be credibly established. The analysis also showed that best price discounts health maintenance organizations (HMOs) and hospital groups received from manufacturers were in fact higher in 1994 compared to 1991 for 30% of the drugs reviewed (Congressional Budget Office, 1996).Footnote 11

Two highly cited studies published in 1997 and 2005 claimed that the Medicaid best price provision resulted in higher costs for the private sector (Scott-Morton, Reference Scott-Morton1997; Duggan and Scott-Morton, Reference Duggan and Scott-Morton2005). The 1997 study concluded that the Medicaid best price provision resulted in a 4% increase across drugstore (but not hospital) non-Medicaid prices for multi-source branded drugs, while it had no significant effect on single-source branded drug prices. The study has severe data limitations and reaches a rather counterintuitive conclusion that the more intense the competition (multi-source drugs) the larger the cost-shifting effect.Footnote 12 More importantly, the impact of OBRA on prices, even if causality is accepted, is, at best, modest: a 4% price increase for a small subgroup of drugs studied. Averaged across all branded drugs, the observed rise in price would be much lower than 4%.

The 2005 study showed that drugs popular with Medicaid tended to be, on average, more expensive, which the authors attributed to OBRA.Footnote 13 However, the observed trend was not statistically significant at the conventional 5% level. Based on the empirical analyses, the Medicaid best customer clause may have led to a very small increase in non-Medicaid prices. Given the methodological and data limitations and the modest effect size, this finding cannot be used to discourage current thinking on a consolidated government approach to purchasing drugs.Footnote 14

Since January 2006, over 6 million Medicaid beneficiaries were moved to the Medicare programme under Part D (Tritz, Reference Tritz2005).Footnote 15 Prices paid by private health plans under Part D are proprietary and cannot be accessed; however, firms’ 10-KFootnote 16 forms show increases in company revenues immediately after the switch (Berenson, Reference Berenson2006; Frank and Newhouse, Reference Frank and Newhouse2007). According to a ‘cost shifting’ argument some of these extra revenues should be channeled back to privately insured employees or other public payers such as VA in the form of reduced drug prices. The limited available evidence suggests the opposite: higher drug prices currently paid by the elderly translate into greater profits for manufacturers.

Dynamic efficiency and the ‘innovation premium’

A consolidated government approach to purchasing drugs equates, for some, to a monopsony situation where the government uses its considerable buying power to negotiate a single (lower) drug price, which could, in turn, reduce industry profitability and future R&D investment. This does not have to be the case. First, it is unclear that the government would choose to lower aggregate payments for drugs. The reform could be introduced in a ‘budget neutral’ fashion as was the case with the Medicare prospective payment system (PPS). There is a broad recognition that drugs play an important role in improving health and that considerable investment is necessary for developing new drugs.

Second, it is unclear how the pharmaceutical industry would respond to less revenue from the federal government. A reduction in revenues could trigger either increased R&D spending to recoup losses, or reduced R&D investment if manufacturers failed to raise external capital. In addition, there is little empirical evidence suggesting that increased revenues from higher prices translate into more R&D expenditure. R&D spending by drug companies as a percentage of net sales has been reasonably constant for the past 20 years despite the higher profits of pharmaceutical companies (Pharmaceutical Research and Manufacturers of America, 2004; Cohen, Reference Cohen2005).

The current financial incentives may not be designed to encourage drug companies to invest in blockbuster drugs. According to Food and Drug Administration data, the number of new drugs continues to fall despite increasing profits (New Molecular Entities slump continues: FDA clears 18 novel drugs in 2006, same as 2005).Footnote 17 Only 14 percent of pharmaceutical revenues is invested into R&D, according to industry-reported data, accounting for less than half of what is spent on marketing (Reinhardt, Reference Reinhardt2001; Gagnon and Lexchin, Reference Gagnon and Lexchin2008).Footnote 18 Perhaps a new pricing system that rewards good value innovation could be designed if the government paid a single price instead of multiple prices that provide mixed signals.

Not all innovation necessarily adds value and not all R&D is necessarily worthwhile from society’s perspective (Dasgupta and Stiglitz, Reference Dasgupta and Stiglitz1980; Tandon, Reference Tandon1983; Skinner et al., Reference Skinner, Staiger and Fisher2006; Dorsey et al., Reference Dorsey, Thompson, Carrasco, de Roulet, Vitticore, Nicholson, Johnston, Holloway and Moses2009; Light, Reference Light2009). Uncontrolled monopoly prices, particularly in the context of price discrimination, coupled with insurance-subsidised demand, can lead to excessive incentives for innovation (Garber et al., Reference Garber, Jones and Romer2006). In fact, the lower the co-insurance rate, the greater the need for some sort of upper bound to price, possibly through negotiating mechanisms, in order to achieve socially beneficial dynamic efficiency.

Finally, prices are not the sole or necessarily the most appropriate means for encouraging innovation. Government already supports drug innovation through intellectual property laws and patent protection, tax breaks for commercial R&D, financing of infrastructure and training of scientists. Most importantly government invests considerable funds in research, particularly at the early and, thus, riskier preclinical stages. For example, the US National Institutes for Health spent over $28 billion in 2007 on medical research.

Above we describe a classical path dependency process, which has led to a lock-in-effect, with the major players resistant to change. The bureaucracy and fragmentation of the administrative system with multiple jurisdictions and congressional committees; an ideological aversion to administered prices or a what might perceived as a move towards a ‘single payer system’; the threat of cost-shifting to the private sector and the possible effects of reduced company revenues on innovation are key obstacles to change and underpin the current opposition to the presidential healthcare reform efforts in the USA.