The effect of canadian imports on prescription drug prices in the u.s.*

The Effect of Canadian Imports
on Prescription Drug Prices in the U.S.*
Warrington College of Business Administration
* This is a revised version of a paper presented at the 2006 Southern Economic
Association Meetings in Charleston SC. Simon Anderson provided many valuable
comments and insights during the preparation of this draft.

I thank Mark Fister for compiling the drug price table and Adam Narkiewicz and Ed See for research assistance. I also thank the Warrington College of Business Administration for financial support of this research. Please do not quote or cite without permission. Abstract
Reimportation of prescription drugs by American consumers from Canada has become a high-visibility policy issue. The large price discrepancies for some patented drugs arise from market pricing in the U.S. and a system of administered pricing in Canada. The model assumes that there are two classes of U.S. consumers: one group who cannot reimport drugs at any cost, and a second group with a distribution of reimportation costs. Under the assumption that the group who can reimport drugs has lower willingness to pay, reimportation serves as a mechanism for price discrimination in The results include the following: 1) a decline in the Canadian price may raise the U.S. price; 2) a shift down in the distribution of reimportation costs may similarly raise the U.S. price; 3) a shift down in the distribution of reimportation costs may raise 1. Introduction
American consumers, health insurers, and health policymakers have become more vocal about their dissatisfaction with high pharmaceutical prices paid by Americans. Many studies have established that there are large discrepancies in wholesale and retail prices between American pharmacies and those in other wealthy countries. Table 1 illustrates differences in mail-order prices between U.S. chains and Canadian firms marketing to U.S. consumers. These price differences are much larger for patented drugs Because pharmaceuticals have high sunk costs for development, prices must exceed marginal costs of production by substantial amounts if drug firms are to continue to develop new drugs. One would certainly expect that drug firms price discriminate across markets whenever they can, and indeed many policymakers have striven to enable drug manufacturers to sell at low prices in poor countries without facing risks of reimportation back to markets in wealthier countries. The price discrepancies between the U.S. and other wealthy countries such as Canada, France, Germany, Japan, and the U.K. cannot simply be the result of drug firms pricing in response to cross-country differences in willingness to pay. Outside the U.S., national health authorities bear a large fraction of pharmaceutical expenditures on behalf of their citizens, and they have implemented a variety of administered pricing systems. While they work in different ways, these pricing systems severely limit the ability of drug 1 Generic drugs may be less expensive in the U.S. than in other developed countries. It should not come as a surprise that generic drugs have a larger share of the market in the U.S. than in countries with administered pricing systems. See McClellan [2003] for evidence on generic pricing and market penetration. firms to earn large monopoly rents on new products in these countries. In contrast in the U.S., managed care systems and pharmacy benefit managers (PBMs) may negotiate prices for their patients, but only some of these organizations have much market power on the buying side. (Ellison and Snyder [2010] explain that prescription volume matters less than the ability of organizations to direct physicians to prescribe particular drugs in a therapeutic class.) During debate on the 2003 Medicare prescription drug plan, proposals to have Medicare negotiate drug prices on behalf of elderly consumers were defeated. One argument in support of market pricing of drugs has been that drug firms must cover their considerable development costs to bring new drugs to market. Mark McClellan, when he was head of the Food and Drug Administration, observed that high drug prices in the U.S. result in American consumers bearing most of the development costs, even though new products benefit consumers around the world (McClellan, [2003]). Many researchers have worried about drug firms concentrating their development efforts on drugs that have high profit potential. Indeed, this is one motivation for proposals to encourage development of drugs needed in the developing world. However, the differences in disease incidence between the U.S. and much of the rich world (especially Canada and Western Europe) are not great enough that research could be directed specifically at the U.S. market alone.2 To some extent, the countries with administered price systems free ride on drug research funded by American 2 What is more distinctive about the U.S. is the market environment for prescription drugs. One reaction to the price discrepancies has been an increase in attempts to import drugs from Canada into the U.S. Since most such drugs are manufactured in the U.S., this is really reimportation. Imports outside of standard manufacturer channels, in particular to frustrate international price discrimination, are often referred to as parallel imports.3 American pharmaceutical manufacturers have fought attempts to import from Canada and have been supported by the U.S. FDA, which is concerned about the feasibility of monitoring safety of parallel imports. Despite this, several state and local governments have announced plans to import drugs from Canada for their employee health plans. Currently, wholesale importation is effectively banned. In contrast, individuals can import products purchased at retail in Canadian pharmacies, although U.S. Customs has began seizing mail-order shipments. In the summer of 2006, Congress considered legislation to legalize imports by individuals [Wall Street Journal, 24 July The debate over allowing imports from Canada has focused on several issues: the effect of reduced profitability on future drug development by private firms; safety issues, and projected reactions by drug firms to exports by Canadian wholesalers. Almost taken for granted in the discussion is that prices will fall, at least on average, in the U.S. We explore the reactions by American manufacturers to increases in imports from Canada in a model with distinct groups of U.S. consumers. 3 Malueg and Schwartz [1994] analyze a model with arbitrage costs and international price discrimination, but no parallel imports occur in equilibrium. Chen and Maskus [2002] and Maskus and Chen [2004] study parallel importing when foreign retailers ship the good to the domestic market. In our model, consumers bear the reimportation costs directly, which allows them to differ across consumers. Pecorino [2002] has studied the effect of parallel imports from Canada in a model with monopoly prices in the U.S. and prices in Canada which are determined through Nash bargaining between the Canadian government and the manufacturers. He shows that Canadian prices will rise if parallel importing becomes widespread. Anis and Wen [1998] discuss the Canadian pricing regime and suggest that a bargaining approach between manufacturers and the Canadian government does not capture the nature of price-setting very well. It is, however, true that manufacturers have threatened to establish quantity limits on Canadian wholesalers if exporting from Canada to the U.S. What Pecorino and other analysts have ignored is the possibility that parallel imports from another country may facilitate price discrimination within a single country. Anderson and Ginsburgh [1999] develop a model of parallel trade where within-country discrimination is impossible, but there are only limited arbitrage possibilities between countries. We use a variant of that model to explore the effects of increases in parallel imports of drugs. There are two groups of American consumers. One group has no ability to engage in parallel importation; consumers with generous health insurance coverage for prescription drugs often have little ability to go outside their health network. Even if they find lower prices (before reimbursement) elsewhere, insurance reimbursement is only available for purchases through specified retailers. The second group of consumers has a distribution of willingness to pay for a prescription drug and a distribution of costs of importation from Canada. These importation costs can be psychic or real (or a combination). If consumers must travel to Canada to get prescriptions filled there, distance to Canada will be an important determinant of reimportation cost. The cost can also reflect some measure of the riskiness of buying drugs outside FDA- approved (and FDA-monitored) channels. Because the first group of consumers have more comprehensive health insurance, the distribution of willingness to pay has a higher We approach these questions through two types of comparative statics exercises. First, since the Canadian price regulation regime appears to have become more stringent over time, we think about the effects of an exogenous change in the Canadian price.4 It is certainly the case that greater spreads between Canadian and U.S. prices have been observed over time and that this has led to increased interest in parallel importing. We also examine the effect of shifts in the distribution in the cost of arbitrage (every consumer’s parallel importing cost changes by the same multiplicative factor). The two effects turn out to be quite similar in many, but not all, respects The main finding is the price charged in the U.S. market (paid by all purchasers in the first group and by those in the second group who purchase in the U.S.) will increase in response to a decrease in the Canadian price or a decrease in the distribution of arbitrage costs. Depending on the level of the Canadian price, the U.S. price may be higher or lower than the U.S. monopoly price in the absence of arbitrage. We also find that profitability may rise or fall as a result of an increase in arbitrage from either source and there increased arbitrage may, but need not, cause the monopolist to want to charge a 4 My analysis differs from that of Anderson and Ginsburgh [1999] in that the Canadian price is not chosen by the monopolist, but rather determined administratively.
2. The simplest model
There are two types of drug consumers in the U.S. Group 1 only purchase drugs in the U.S.—because their cost of importing drugs from Canada is prohibitively high. One possibility is that this group consists of consumers with private health insurance that covers prescription drugs. Their insurance plans only reimburse drug purchases through the plan, so they have no incentive to purchase off-plan. Their demand curve is Q1 = a – P1, where we should interpret P1 as the full price paid by the insurer and the insured for The second group of consumers may purchase drugs in the U.S. or in Canada (or through any other outlet, but we can think of Canada as the sole alternative). Let  = fraction of group 2 consumers who buy in the US (1-  of the group 2 consumers purchase from Canada). The U.S. demand from these consumers is Q2 = (b – P2), or . For this simple model, assume that the two groups of consumers are equal in size—we will relax this assumption in the more detailed model of In principle, the division of group 2 consumers into Canada and U.S. purchasers will depend on the U.S. price, the Canadian price, and the distribution of parallel import costs of consumers. Holding the fraction buying in Canada fixed to determine individual drug prices is appropriate if the decision to shop in Canada depends on prices in the aggregate (as in an insurer’s decision to import Canadian drugs). 5 This insured group would also never purchase drugs in Canada unless the retail price in Canada were lower than the net price after reimbursement in the U.S. First, we can find the optimal single monopoly price as a function of  and show that it has some desirable properties for our model. Add the U.S. demand curves together to obtain: Q  a   b  (1  )P . Q . Let marginal cost be constant and equal to c. Setting MR = MC, the non-discriminating monopoly price is the solution to: a   b  1   c a   b  1   c For reference, the discriminating prices in the U.S. would be: The quantities sold at this price to the two groups are: Thus, a decrease in  (a rise in Canadian imports) increases the monopoly price in the U.S. if b < a (if the market segment which is sensitive to the Canadian price has the lower willingness to pay—and thus has more elastic demand at a uniform price). We could easily make the parallel import cost linear, with a uniform distribution of these costs. That is, consumers would import the drug if P* > PC + t, where t is uniformly distributed among group 2 consumers and PC is the Canadian price. If drug producers were to take  as given (even though they are monopolists in their product, the import decision depends on the aggregate prices of a large bundle of drugs in the US and Canada), we can derive (P*, PC, F(t)) and solve for similar comparative statics with respect to PC and F(t). In the next section, we examine a model where each producer takes account of the effect of its own price on the level of reimportation.
3. A more sophisticated model
Initially, we will take the Canadian price (denoted by PC) as given. Holding the Canadian price fixed is intended to reflect the fact that drug prices in Canada are set by an administrative procedure (Anis and Wen, 1998). As before, there are two groups of US consumers. Group 1 has no ability to buy drugs in Canada; the consumers have a uniform distribution of reservation prices (denoted by r) from 0 to a and there are N1 of this type. Thus, demand from group 1 consumers equals D1 = N There are N2 consumers in group 2, and they are defined by two characteristics: their reservation price and their cost of importing from Canada. Let denote a consumer’s cost of importing, where t is uniformly distributed from 0 to . The reservation prices are distributed uniformly from 0 to b, and reservation prices are independent of import costs. If distance from the Canadian border is the primary influence on importation costs (as when consumers must travel to Canada for purchase), independence of willingness to pay and importation cost seems reasonable. Group 2 consumers purchase in the US if r  P and t  α(P  P ) , where P Canadian price. The parameter  allows us to examine shifts in the distribution of transport costs. Group 2 consumers buy in Canada if r  PC + and t < α(P  P ) . We for purchases in Canada. See Figure 1 for a graph of these regions in (r, ) space. In the absence of parallel imports, the discriminating prices in the US would be: and the nondiscriminating monopoly price is a weighted sum of these two: Let c denote marginal cost. Then we can write profits as the sum of profits from To find the profit-maximizing price (assuming no ability to discriminate within the US), taking as given the Canadian price (set using a reference price system), we a  c  2P Pc  P P  b P   P P  b P Pc 2b  P  P P cP  P P c N a  c  2P 2bc2P  P P 2 b PPc2 b PP c b  c  2P  P  P  b  P P  P . We assume throughout that parameter values are such that we are not at a corner solution where only group 1 consumers buy in the U.S. There does not appear to be an We can consider the comparative statics questions of what happens as the Canadian price or the distribution of transport costs shifts. Totally differentiating the FOC, we obtain:  0 from maximization, dP has the same sign as b  c  2P  b  P = 2 2b  c  3P .  0 , and a decline in the Canadian price (perhaps as the result of tougher negotiation by the Canadian single payer) will cause an increase in the in the absence of imports, this is a higher threshold, so that this condition may or may not hold. , where an increase in  corresponds to a decline in transport cost for all group 2 consumers. We find that: bc2P(P  P )(b P)(P  P ) 2b  c 3P, which always takes the opposite sign of  0 . More consumers buying in Canada (for either reason) moves the domestic price in the same direction.  equals zero, we can solve for the equilibrium price, which is the same as in the absence of parallel imports. Evaluating 2b  c  3P at  = 0, we obtain: b  c  3P  0 if 2b  c  3  0  3 a  b  3 b  c , . (At   ,  , for example, and a can be smaller for larger values of c.) When this holds true, dP  0 at   0 , and thus dP  0 and dP  0 for higher values of . There are a few interesting bounds on the US price to look at for particular values of the Canadian price, PC. This will indicate some of the incentives for the producer and help with some intuition. First, if the Canadian price equals marginal cost (an extreme a  c  2P   P cb  c  2P b  PP c . (the nondiscriminatory monopoly price in the        1    P cba b  PP c  b  ab  P P  c b c 3 a b (the monopoly price when no US consumers buy in  0 , and the arbitrage possibilities reduce the price in the U.S. For a Canadian price this low, the monopolist wants type 2 consumers to purchase in the U.S. a  c  2P 2 b  c  2P  b  P P  1 a b  bc3 a b  b  c 3 a b    1 a  b  1  b  a Thus, if b  c  3 a  b ,  0 at P*, and the Canadian imports result in the U.S. price exceeding the monopoly price in the absence of arbitrage. Since b  c  a  b must hold for both types to buy at the monopoly price, the full requirement is that a  b  b  c  3 a  b . In this case, the Canadian price is high enough that losing sales to Canada helps the monopolist price discriminate between type 1 and type 2 consumers. 4. Endogenizing the Canadian price
We have focused on the feedback effects of changes in the Canadian price on the US market, while viewing the Canadian price as being determined in an administrative process. Pecorino [2002] has studied a Nash bargaining process for the Canadian price and how that changes with the development of a parallel import market. It is worth examining how the monopolist’s Canadian price would change with the development of parallel imports. (This analysis may be more relevant for imports from developing countries—such as Mexico—where the monopoly price may be significantly lower than in the US. It seems hard to explain the US-Canadian price differential by differences between US and Canadian consumer incomes and preferences.)         (P )   P, P   (P ) denote total profits in the two markets. In the absence of parallel imports, d C  0 determines the Canadian price. With parallel imports, the FOC changes to: b  PP c  0 and 2b  P  P P  c  P  P  P P c 3P  4bP  P  2bP  2cP  2  3P  4bP  3P  2cP  2cP  4bP . The term in braces can be written as: P  P Since b  P  c , this expression is positive for PC near c, so that parallel imports may put upward pressure on the Canadian price. If PC = price), this expression is positive for P 
5. Effects on profits
We can also examine the effect of a change in the Canadian price or in the cost of       , we can use the envelope b  PP  c b P b PP cP  P P cP  P b P b P 3P  4bP  3P  2cP  2cP  4bP . The ambiguities in this expression were discussed above. The effect of a change in the cost of arbitrage is: b  PP  P P c P  P b  P b  PP c . An increase in  corresponds to easier arbitrage—for the same cost difference, there will  b  PP  c  b  P  b  P P  c . If c = 0, the expression in braces equals 2b  P P  2b  P  P P . This takes on its  b  P P   b  P  P P   bP  P  b  bP  P  9P 12bP  4b   3P  2b2 Hence, there is a range of values for PC and c such that that making arbitrage easier can raise profits from US consumers. American manufacturers have opposed allowing consumers to import patented drugs from Canada and have threatened to impose restrictions on Canadian wholesalers who export. If American manufacturers oppose allowing imports, presumably they do not think that reimportation would increase their profits. In order for easier arbitrage to raise manufacturer profits, the Canadian price needs to be sufficiently high. For low Canadian prices, manufacturers would lose from easier arbitrage.
6. Conclusions and further work
There are a number of qualifications to these results, even in this stripped-down model. First, unlike some other instances of parallel importing, the domestic manufacturers of prescription drugs may have considerable opportunity to price discriminate within the domestic market. Whether they can discriminate between the two groups in our model (low or high reimportation costs) is less clear. The high importation cost group includes many consumers with insurance coverage for prescription drugs. The low-demand group may correspond to consumers with no prescription drug coverage— they may have the least bargaining power and not be in a position to benefit from Currently, consumers covered by the Medicare prescription drug plan cannot gain from bargaining for favorable prices. Some insured consumers obtain prescriptions through pharmacy benefit managers (McGahan [1994]) who do negotiate prices with manufacturers. Many other insured consumers do not benefit from reduced wholesale prices because of a lack of bargaining (Ellison and Snyder [2010]). If groups of insured consumers benefit from price discrimination, our model can be thought of as describing the residual market of consumers who do not buy at negotiated prices. If threats of Canadian imports permit groups of insured consumers to negotiate lower prices, our model could accommodate that by letting the sizes of the two groups buying at the standard US price adjust as reimportation became easier. One must consider that this does not appear to be a typical environment with price discrimination in which a group with lower willingness to pay and more elastic demand obtains a lower price than another group. The consumers who obtain reduced prices for prescription drugs are those who enroll in insurance programs that can do a better job of keeping physicians adhering to formularies. If consumers in this group could import drugs from Canada as only individual purchasers (in contrast to through their health plans), reimportation possibilities and the US-Canadian price differential seem unrelated to the prices these groups can negotiate.
References

Anderson, S., and V. Ginsburgh, Price discrimination with costly consumer arbitrage,
Review of International Economics 7: 126-139 (1999). Anis, A., and Q. Wen, Price regulation of pharmaceuticals in Canada, Journal of Health Chen, Y., and K. Maskus, Parallel imports in a model of vertical distribution: Theory, evidence, and policy, Pacific Economic Review 7: 319-334 (2002) Danzon, P., and L.-W. Chao, Cross-national price differences for pharmaceuticals: How large, and why?, Journal of Health Economics 19: 159-195 (2000) Ellison, S., and C. Snyder, Countervailing Power in Wholesale Pharmaceuticals, Journal of Industrial Economics 58: 32-53 (2010) Malueg, D. and M. Schwartz, Parallel imports, demand dispersion, and international price discrimination, Journal of International Economics 37: 167-195 (1994) Maskus, K., and Y. Chen, Vertical price control and parallel imports: Theory and Evidence, Review of International Economics 12: 551-570 (2004) McClellan, M. Speech before 1st International Colloquium on Generic Medicine, U.S. FDA, http://www.fda.gov/NewsEvents/Speeches/ucm053614.htm (2003) McGahan, A. Focus on Pharmaceuticals: Industry Structure and Competitive Advantage, Harvard Business Review, pp. 115-124, Nov-Dec (1994) Pecorino, P., Should the U.S. allow prescription drug reimports from Canada?, Journal of Health Economics 21: 699-708 (2002) Mail-Order Prices for Selected Prescription Drugs (See endnotes for data description and sources.) Min CDN price
Max CDN price
Min USA price
Min USA/Max CDN
Min USA/Min CDN
Lipitor
10mg
Lexapro
10mg
Nexium
20mg
Singulair
4mg
Plavix 75mg
Advair Diskus
100/50mcg
Effexor XR
37.5mg
Protonix
20mg
Lotrel
5mg/20mg
Allegra 60mg
Diovan
40mg
Wellbutrin XL
150mg
Celebrex
100mg
Zetia 10mg
Avandia
2mg
Viagra
25mg
(transport
no purchase
t
α P − P )
r
(reservation price)

Source: http://warrington.ufl.edu/departments/eco/docs/papers/Hamilton_DrugPrices.pdf