• Have any questions?
  • info.zbook.org@gmail.com

10plus2minus5plus36plus3minus34plus2minus8plus2minus44plus .

3m ago
1.86 MB
36 Pages
Last View : Today
Last Download : n/a
Upload by : Matteo Vollmer

Markups and Fixed Costs inGeneric and Off-Patent Pharmaceutical Markets Sharat GanapatiRebecca McKibbinGeorgetown UniversityThe University of SydneyMarch 2021AbstractPharmaceutical prices are widely dispersed across countries with comparable quality standards. Under monopoly, off-patent and generic drug prices are at least four times higher in the United Statesthan comparable English-speaking high income countries. With five or more competitors, off-patentdrug prices are similar or lower. Our analysis shows that differential US markups are largely driven bygeneric supplier market power and not wholesale intermediaries or pharmacies. Furthermore, we showthat the traditional mechanism of reducing market power, free entry, is limited as implied entry costsare substantially higher in the US.JEL Codes: I11, F14, L44 Thanksto Joe Shapiro, Chris Snyder, Steve Berry, Fiona Scott Morton, Jason Abaluck, Jon Skinner, Leila Agha, JohnRust, Nina Pavcnik, Stuart Craig, Samuel Moy and seminar participants. Thanks to Serena Sampler and Ray McCormick forexcellent research assistance. Part of this research was conducted while Ganapati was a Fellow at Dartmouth College and atLudwig-Maximilians-Universität Munich. All errors are our own.Contact information: Sharat Ganapati - sg1390@georgetown.edu, and Rebecca McKibbin - Rebecca.mckibbin@sydney.edu.au.1

1IntroductionWould increasing competition reduce the prices of off-patent pharmaceuticals in the United States? TheUS largely relies on competitive forces to set the price of pharmaceuticals that are no longer protected fromcompetition by patents (primarily, but not exclusively, generic drugs). It is assumed that once patentsexpire, generic versions of the drugs will enter and drive prices down to competitive levels. In contrast, manycountries with comparable safety standards use a combination of competition and government purchasingpower to attain low generic drug prices. We document that US generic and off-patent-drug’s prices inmarkets with few competitors are higher than in these comparable countries. We then investigate theseprice differences to understand the relative roles of competition between pharmaceutical suppliers that facebarriers to market entry versus markup along the value chain.In the US, generic drugs account for 90% of prescriptions and 23.2% of expenditures in the 324 billionprescription drug market (IQVIA, 2018). Empirical evidence indicates that off-patent drugs are not soldin perfectly competitive marketplaces at marginal cost. For example, the price of pyrimethamine, an antiparasitic developed in the 1950’s, was changed overnight from 13 to 750 per dose (Pollack, 2015).1 Althoughno longer patent protected, many markets may not be large enough to attract the number of competitorsneeded to achieve marginal cost pricing.2 They may also be amenable to collusive behavior (Rowland, 2018).Motivated by these observations, we compare final, wholesaler, and manufacturer/supplier prices forpharmacy dispensed off-patent drugs across the US, Australia, Canada, New Zealand, and the UK.3 Thisreveals that the final price of off-patent drugs in the US relative to each of these other countries declines asmore suppliers enter the market.4 With one American supplier, the price at least four times higher. However,when there are five or more suppliers, retail prices are either similar or lower. We focus on prices paid bythe US Medicaid program, however, results for Medicare Part D and private insurance are broadly similar.We additionally compute the price premium using prices that approximately capture relationships atearlier points in the supply chain: specifically the price paid to manufacturers and the acquisition costfor pharmacies. This suggests that the identified price premium is driven by manufacturers rather thanintermediary pharmacies, wholesalers, or benefit managers, and that the current price mechanism used byMedicaid programs is not an effective tool in small markets. It does not appear that there is sufficientrevenue available to support the fixed costs involved in additional entry in these small markets. Our resultssuggest that while competition is an effective mechanism for lowering prices in markets with many suppliers1 Since2010, 20% of US generic molecules have temporarily doubled in price (GAO, 2016).et al. (2017) and Dave et al. (2017) show that 50% of US markets are monopolies or duopolies.3 For Canada we focus on the two largest English speaking provinces: Ontario and British Columbia. For the UK, we solelyuse England.4 While the US has higher prices for drugs with limited competition, it often has the lowest prices for the most widelydisseminated drugs.2 Berndt2

and manufacturers, which tend to be large markets with many patients, smaller markets have high fixedcosts preventing entry and competition. We find that fixed costs must be four times higher in the US thanAustralia. If implicit fixed costs cannot be lowered, some form of second best price intervention such asbargaining may be necessary.We build on Berndt et al. (2017), who described the landscape of generic drug markets in the US andnoted the potential absence of competition, and a previous empirical literature that makes cross-countrycomparisons of drug prices (Danzon and Furukawa, 2003, 2011; Kanavos et al., 2013; Wagner and McCarthy,2004; Danzon and Chao, 2000a). We build a uniquely detailed dataset using publicly available prices inorder to first examine whether competition is indeed failing in some markets in the US, and then investigatewhere along the supply chain this occurs. Previous cross-country drug comparisons have focused primarilyon determining which country (or system) produces the lowest prices. Danzon and Chao (2000b) broadlyconsider if one system of regulation has systematically lower or higher prices (using data from the 1990s),we seek to understand the the current situation the US, which has changed considerable in the past thirtyyears (Berndt et al., 2017), and how and why it varies in different segments of the market - as well as upand down the value chain. This paper also contributes to the broader literature on generic drugs, which asprimarily focused on the effect of exogenous entry on list prices (Reiffen and Ward, 2005; Berndt et al., 2017;Grabowski and Vernon, 1992), or how competition-based policies could increase competition and hence lowerprices (Scott Morton, 1999; Berndt et al., 2017; Gupta et al., 2018; Bollyky and Kesselheim, 2017; Berndtet al., 2018). In our analysis of fixed costs, we incorporate an empirical literature studies the substitutionpatterns between generic entrants and the original branded drug (Kanavos et al., 2008; Puig-Junoy, 2010;Kanavos et al., 2013).2Institutional BackgroundCompetition in off-patent drug markets is shaped by the nature of manufacturing supplier approvals, theinsurance system, and pharmacy regulation. We briefly describe how these elements affect prices at eachstage of the supply chain in each country as it applies to generic and off patent patient-administered drugspurchased in pharmacies. At a high level, the supply chains share many similarities. In all five countriessuppliers or importers obtain approval from a regulatory body to market their products, the suppliers sell towholesalers, who sell to pharmacies, who retail the products to consumers. The retail price paid depends onthe consumer’s health insurance policy. However, the US differs in an important way - there are many insurers(or their subcontracted pharmacy benefit managers) arranging their own price agreements with pharmaciesand suppliers (with potential consumer cost-sharing). In contrast, other countries have one predominantprice setter - the government, which enables them to offer a set price schedule for generic drugs.3

2.1Entry RegulationsManufacturing and entry regulations are nearly identical across our sample countries. Before a drug canbe sold, the supplier must apply for local regulatory authorization. These regulators monitor quality thedrug’s safety, efficacy, and manufacturing quality. Approval requirements depend on whether the drug is aninnovative drug or a generic. Innovative drugs require extensive (and costly) clinical trials to demonstratetheir safety and efficacy profile. Generic approval requires demonstration that the generic is bioequivalentto an innovative drug. Bioequivalence typically requires a clinical trial that shows the generic delivers thesame amount of active ingredients in the same amount of time as the original drug. The generic must also beidentical in terms of dosage form, strength, route of administration, and intended use. All five countries followthe International Conference on Harmonization (ICH) guidelines in assessing bioequivalence. In addition todemonstrating therapeutic equivalence, suppliers must also demonstrate compliance with production qualitystandards. All five countries have Good Manufacturing Practice (GMP) guidelines that comply with theICH. These regulations ensure that products are properly produced, packaged, and safe (FDA, 2018).2.2Insurance & PricingFinal prices are determined by the relationships between wholesalers, pharmacies, insurers, and manufacturers. The US has three major insurance regimes covering prescription drugs sold in pharmacies: privateinsurance, Medicare Part D, and Medicaid. In 2016, private insurance accounted for 41% of retail prescription drug spending, while Medicare and Medicaid accounted for for 30% and 11%, respectively (theremainder is out of pocket payments (15%) and other government programs (4%))(CMS, 2018). However,within these three groups there are numerous independent insurers - Medicaid is itself is a separate programin each state, and each state may be have a component outsourced to private managed care plans. Allthree types of insurance frequently outsource the management of their prescription drug plans to pharmacybenefit managers (PBMs). PBMs are concentrated, with three companies controlling 66% of the market(Sood et al., 2017). PBMs act as intermediaries between health plans, pharmacies, and suppliers, as well asset the formularies and network of covered pharmacies. For drugs with a single supplier, PBMs negotiatedirectly with suppliers for favorable placement on formularies (when there are drugs in the same therapeuticclass that are close substitutes) in return for rebates, some or all of which they keep as profit. When thereare multiple suppliers, PBMs provide a Maximum Allowable Cost (MAC) list to pharmacies, which stateshow much they will pay for the drugs. While MAC lists for private insurers and Part D are between theinsurer and the PBM, Medicaid reimbursement rates are subject to statutory rules. These vary by state,but broadly, they compute potential prices using four different methods and set the reimbursement at the4

the lowest one. In addition, a federally mandated manufacturer rebate is set by statute.5In contrast with the US, Australia, the England, and New Zealand have government-operated universalhealth insurance plans while Canada has a mixed public/private system.6 In all of these countries, thegovernment uses its purchasing power to obtain lower drug prices. For on-patent drugs, governments directlybargain with suppliers. In the case of generics, each country stipulates how it will pay and the suppliers cantake it or leave it. The method used for setting generic drug prices differs across countries. Australia andthe England both use a reference price system, based on reported supplier prices. Suppliers report their salesprices and the health plan reimburses at the average. In Australia, prices are not revised upwards withoutapproval, whereas in the England, the price fluctuates with the market. Australia also has a mandatoryprice reduction of 16% when the first generic enters the market. New Zealand uses a competitive tenderingsystem to obtain low generic drug prices. The supplier of the winning tender has their product exclusivelyeligible for reimbursement by the national health plan. All Canadian provinces except Quebec have agreedto set the price of generics entering the market from 2014 onwards using a tiered pricing system.7All five countries have two further intermediaries between the consumer and the manufacturer: wholesalers and pharmacies. The wholesale market is very concentrated in all five countriesAustralia, theUS, the England, and Canada have three dominant wholesalers, while NZ has just two wholesalers. Thepharmacy landscape varies with Canada and the England being similar to the US, with large retail chainsdominating the market.8 In Australia and New Zealand, there are regulations around pharmacy ownership,which make them more difficult to compare.93FrameworkOur empirical analysis can be understood within the framework of a static entry model (in the spirit ofBresnahan and Reiss (1991) and Scott Morton (1999), which are the appropriate choice for stable marketsfor long off-patent molecules). Generic suppliers enter the market in destination f and provide product dif the profit of doing so (πf,d ) is greater than the fixed cost of entry Ff . We assume that there is a fixedcost of entry into each destination (Ff ) that is destination specific and independent across destinations.10 A5 Duringour study period this rebate was a uniform 13% for generic drugs.may have private insurance and each province has its own public drug plan, which cover disadvantaged segmentsof the population. We focus on the two largest English-speaking provinces: Ontario and British Columbia.7 The reimbursement rate is set at 85% of the price of branded drug when there is one generic, 50% with two generics and25% when there are three or more. For generics introduced before 2014 the provincial reimbursement rates are used. Ontariosets the reimbursement rate at 25% while British Columbia sets it at 20%.8 The largest five US pharmacies account for approximately 64% of the market (Drug Channels Institute, 2019), in theEngland they hold 80% (Sukkar, 2016) and in Canada 60% (Yeates and Hernandez, 2019).9 In Australia no entity can own more than five pharmacies (Hattingh, 2011), but four corporations control 63% of the marketthrough franchising (Medicine.com.au, 2019). In New Zealand, a pharmacy must be owned by a pharmacist. However, 45% ofpharmacies have some affiliation with a single corporation (Pharmaceutical Society of New Zealand Incorporated, 2017)10 The second part of this assumption is less restrictive than it initially seems. We consider marginal generic entrants - asopposed to the original innovator. Many of the implied fixed costs are country-specific, such as the regulations, lobbying,6 Workers5

marginal supplier s enters a destination d if and only if:πf,d (s; S) [µf,d (s; S) 1] cf,d qf,d (s; S) FfThe profitability (πf,d ) of the marginal sth supplier over the set of S suppliers of entering each market is theproduct of a destination specific mark-up µf,d over marginal cost cf,d , multiplied by the quantity sold qf,d .We assume that the marginal cost within a market in a destination is constant across suppliers (cf,d,s cf,d ), as approval requires the same raw materials.Next we consider the price. As described in Section 2, there are many actors that are involved insetting the price of a product along the supply chain. These actors include wholesalers, pharmacy benefitmanagers, pharmacies, consumers, insurers and governments. We focus on the retail price of each market(molecule-dose-route) in each destination averaged across products pf,d . The retail price p of a productcomprises an amount paid by consumers (a copay δcopay ) and the amount the government contributes (δgov ): p δgov δcopay . The retail price ps,f,d of a product d in destination f can be represented as as as a seriesof mark-ups (µs,f,d ) over the common marginal cost of producing the product:BMwholesalerps,f,d µpharmacy µP µsupplier cf,d .s,f,d µs,f,ds,f,ds,f,dWe choose to use the entire retail price rather than just the government contribution as the governmentmay subsidize markets for redistribution purposes and so we view this as the welfare relevant object. We takethe average of the retail price across products because we view changes in the average price as best capturingwhat is ultimately paid. Retail prices differ across pharmacies as individual consumers select the pharmacyto fill their prescription. There is generally no restriction on pharmacy choice in the non-US countries, butlimited restrictions in the US.The price we measure is therefore:pf,d Xshare (s) ps,f,d .s SThis market weighted average share allows for differential pricing (and markup) between different suppliers.In some markets, there is substantially different pricing and market shares between the original brandedentrant and other generic entrants (Kanavos et al., 2008). We account for this when we consider the boundfor a fixed cost to enter a market and gain distribution:Ps S µd,m,s qd,m,s Is6 BrandedP Fm ,s S Is6 Brandedwhere we simply focus on marginal, as opposed to the original branded entrant (e,g, generic atorvastatinbilateral payments, and/or campaign contributions. Physical manufacturers are often “contract” manufacturers, which areseparate from the official government-approved supplier. Contract manufacturers receive a specification and produce finalpackaged products for the government-approved supplier and simply have to pass safety related inspection measures and arenot responsible for the vast majority of entry costs, which are borne by the government approved supplier (Miller, 2017). Lastly,a survey of foreign factories for a sample of drugs shows that there are a large mass of potential suppliers that do not appearin our sample. For example, there are only 10 FDA approved US suppliers for acyclovir, but at least 148 brands approved forsale somewhere(MedIndia, 2019). See the appendix for a comprehensive analysis.6

versus Pfizer’s Lipitor).4DataWe compare prices, competition structure, and market size across countries for off-patent patient administered drugs.11 This requires data on the prices, quantities of prescriptions, and the number of suppliers ofeach drug for all five countries. We define a drug market as all products with the same molecule-dose-route.12This follows the definition used by the US Food and Drug Administration (FDA) in determining genericentry eligibility. We allow for imperfect substitutability between branded and unbranded generics, whichhas been shown in the literature to be an important distinction (for a summary see Kanavos et al., 2008).Data on molecule-dose-routes in the US are obtained from the drugs@FDA database.We includemolecules administered as “capsule” or “tablet”. We only include markets where the original product isoff-patent and hence generics can enter (regardless of whether they have). A molecule-dose-route is classifiedas off-patent if an Abbreviated New Drug Application (ANDA) has been approved or if there are no patentsor exclusivities listed in the FDA orange book. We compute similar statistics at the molecule level. As ittakes time for the first generic to be reviewed, and the first approved generic receives 180 days of exclusivityagainst additional generic entry, we exclude markets with the first FDA approval within 20 years.13 Thenumber of suppliers supplying each included US market reflects the number of approved ANDAs. Data ondrugs for corresponding foreign markets are obtained from the relevant regulatory authorities. We mergeeach of these data sets together and keep the subset of drugs that are available in both the US and the othercountry. Details in Appendix A.The retail price and quantity prescribed in each off-patent market in the US is obtained from the MedicaidState Utilization data, which we aggregate across states. We compute per-unit prices net of dispensing feesand manufacturer rebates.14 We focus the analysis on Medicaid because it is the only publicly availablesource of prices that can be adjusted for manufacturer and pharmacy dispensing fees, which is needed inorder to make the US data directly comparable with international data. As Medicaid only accounts forabout 10% of US prescription drug spending, we include data from Medicare Part D and private insurance11 Wemix the usage of the term generic and off-patent, even though not all off-patent drugs are the generic copy.robustness, we also conduct the analysis with markets defined at the molecule level. We also examine the potential roleof therapeutic substitutes as defined by the Anatomical Therapeutic Chemical (ATC) Classification codes.13 This is similar, but not identical to patent protection, a period that typically starts at discovery, but before clinical trialsand FDA approval, a process that can take upwards of 10 years. The US Hatch-Waxman Act effectively allows for five years ofexclusivity after the patent expiry, allowing for 25 total years of near exclusivity. Our window omits drugs in this period.14 Drug prices in the US are difficult to measure. There are many different types of drug prices available, many of whichare not true prices but rather list prices. These adjusted prices for Medicaid account for rebates and payments from themanufacturer back to the government. We also obtain data on ex-manufacturer prices, which is the price received by thesupplier. Manufacturer rebates cannot be directly computed for all drugs in our sample, which is set at 13% of the AverageManufacturer Price. For those missing data points, we substitute 13% of the final retail Medicare price as an upper bound.Broadly similar subsample results are available in the appendix.12 For7

(through National Average Drug Acquisition Cost dataset), as a robustness check. These datasets and theanalysis are described in the Appendix.Data on retail drug prices and ex-manufacturer prices in Australia, Ontario, British Columbia, England,and New Zealand are obtained from the national health plan administrative statistics. We convert all pricesinto US dollars using the average annual exchange rate for each calendar year. We recover the per-unitprice of a drug, net of a fixed per-prescription pharmacy dispensing charge. For Australia, we use the pricepaid by Pharmaceutical Benefits Scheme (PBS) to the pharmacy, available from Schedule of Benefits for2009-2017. This price excludes the dispensing fee and the allowed pharmacy mark-up.15 For Ontario, we usethe Ministry of Health and Long-Term Care’s drug benefit prices 2017. This price omits the fixed dispensingfee and variable patient copayment, which can only offset the dispensing fee. Data on British Columbia fromBritish Columbia PharmaCare for 2014-2017. We use the maximum allowable price, the maximum amountthe drug benefit will reimburse, net of dispensing fees.16 Any difference between the retail price and thisprice is paid by the patient. For New Zealand, we obtain data on retail prices from PHARMAC. This priceexcludes the patient copay, which we add to reported price. Prices for England are obtained from the NHSEngland Drug Tariff. These prices are the amount of the NHS subsidy. It does not include professional feespaid to pharmacists for dispensing the products.Table 1 shows the number of observations for each country and the years included in the analysis. Pricesfor all non-US markets are net of fixed per-prescription pharmacy dispensing fees. The key variable of interestis the relative price of each drug in the US compared with the same drug in each of the other countries.The mean and variance of the US price ratio, with respect to each base country are shown in the fifth andsixth columns of Table 1. On average, prices are extremely similar, with US prices only slightly higher thanforeign prices. Adjusted prices subtract out the average pharmacy dispensing fee as well as the statutorymanufacturer rebate of 13% for non-innovator drugs.Table 1 highlights the variation in suppliers in the US for our sample. While drugs with just one supplieraccount for 1% of doses, they make up 10% of off-patent Medicaid spending. Drugs with five or fewersuppliers account 25% of doses, but for 50% of total spending. While the majority of doses sold are incompetitive markets, likely priced near marginal cost, many drugs have a limited number of suppliers.15 Patient copays in Australia are variable and are capped at AU30.70- AU38.80 during this period. Generic drugs oftenhave patient copays below the cap, as maximum co-pays are capped at the combined cost of the pharmaceutical, dispensing,and preparation fees. Reported prices are inclusive of this variable patient copay.16 Which are currently capped at C10.8

5AnalysisWe compare the retail price of off-patent molecules across the destinations and show that in markets withfew domestic suppliers, Medicaid pays substantially higher prices than foreign governments. We then showevidence that suggests the price premium is due to suppliers, as opposed to pharmacy or wholesaler, markups,which dissipate as the number of domestic suppliers increases.5.1Prices are higher in markets with low competitionWe compute the relative price differential between the US Medicaid price and a foreign price as:priceU Sdypremiumf dy ,pricef dy(1)in molecule-dose-route combination d, year y, in comparison to foreign market f . The premium allows for thedifferencing out of multiplicative market-level fixed effects between country pairs, netting out the commoncost of manufacturing. We recover the relationship between the premium and the number of suppliersparticipating in the US marketplace with US FDA regulatory approvals to distribute drugs: Xln premiumf dy βs IS (SU Sdy ) δf y f dy .(2)IS is an indicator function for the number of US approved suppliers in year y in market d: SdyU S . βs are avector of prices differences, δf y are year-comparison market fixed effects, and f dy represents measurementerror.Figure 1 shows the βs from Equation 2 of the premium paid by Medicaid relative to each foreign countryby the number of US suppliers in 2017.17 As the number of suppliers increases, the premium declines. Witha single supplier in the US, the price is at least 400% higher than in comparison destinations. The pricepremium declines as there are more suppliers in the US market. US drugs are cheaper when the US has 5or more suppliers. This pattern holds across all comparisons and for all various types of US insurance (seeAppendix B using other insurance types, including private data and Medicare).These results are formalized by estimating regressions of the price differential between US Medicaid andanother country on the of the number of US suppliers . Yearly fixed effects capture exchange rate fluctuations. ln premiumf dy β ln (SU Sdy ) δf y f dy .(3)As is standard in demand estimation, we assume prices are more flexible than market entry and have nodynamic effects. If represents a mean zero shock after supplier entry, we can interpret β as a causalrelationship between the number of suppliers and the pricing differences (but not absolute price levels).Table 2 shows the results. A one percent increase in the number of suppliers in the US is correlated with17 2017 is the only year with data for all comparisons. Pooling data across years, and estimating the coefficients relative tothe 7 category yields a similar pattern but has less clear interpretation because the estimates are relative to the base. Similarresults hold for the NADAC data and for Medicare Part D. See Appendix section B.9

price differential decrease between US Medicaid and Australia by 1.2%, British Columbia by 1.2%; NewZealand by 0.8%; Ontario by 1.0%, and England by 0.90%.There are a few clear threats to identification, from both the supply side and the demand side. Onthe supply side, may allow for differences in marginal cost between markets. Marginal costs may besystematically related to the number of suppliers. However, we directly control for the marginal cost ofproduction by considering the relative prices between two nations and absorb differences in distributioncosts and exchange rates using the country-year fixed effect δf y .On the demand side, there may be differences in substitutability between markets. For example, theremay be many cardiovascular over-the-counter alternatives, but very few such anti-epileptic alternatives.However, our relative price differences are robust, controlling for the age of a drug, the number of similardrugs, Anatomical Therapeutic Chemical (ATC) code-year fixed effects, patented drugs within the ATCcategory, the number of available dosage forms, and using the lagged number of suppliers to control forsticky prices.18 These relationships have been stable since 2010, with limited observable changes in therelationship between suppliers and price premiums. We now take a look at competition in foreign markets.It is possible that the price differential can be explained by differences in the market size across destinations, due to differences in preferences for new versus generic drugs or differences in the prevalence ofdiseases. We examine this issue by estimating Equation 4, which controls for the number of suppliers in thecomparison destination. Assuming marginal cost are the same across countries: ln premiumf dy βU S ln (Sf dy ) βF ln (Sf dy ) δf y f dy .(4)If the price premium can be explained by the foreign markets having more suppliers than the same USmarket, controlling for the foreign supplies should explain the price premium. If not, this suggests a largerrole for differences in market structure and markup determinations. Panel B in Table 2 shows results withboth the number of US and comparison country suppliers. Foreign suppliers are insignificant for threecomparisons (Ontario, New Zealand, and England). A change in the number of competitors in a foreignmarket ar

Jun 04, 2018 · MarkupsandFixedCostsin GenericandOff-PatentPharmaceuticalMarkets Sharat Ganapati Georgetown Univer