Non-Tariff Barriers and Bargaining in Generic and Off-Patent Pharmaceuticals Sharat Ganapati - Georgetown University Rebecca McKibbin - University of Sydney May 2019 Ganapati+McKibbin 1
Motivation Some off-patent drugs are expensive in Other off-patent aren’t expensive the US US Price: Gabapentin: $0.17 -20 Mfg UK Price: Gabapentin: $0.24 -11 Mfg US Price: Pyrimethamine: $750 - 1 Mfg UK Price: Pyrimethamine: $10 -1 Mfg Globally, > 76 Manufacturers Ganapati+McKibbin 2
Two questions Policy Question : How to set prices for older drugs? » The US (mostly) leaves drug pricing to the market » After patent expiration, generic entry should drive prices down Entry may take time, but there are persistent differences across countries. Economic Question : Why does the law of one price not hold? » Possible reasons: Trade Barriers? Must be fixed cost based; low transport costs + 0 tariffs. Imperfect Competition? Due to entry costs/trade barriers? Or are markets sizes naturally limited? Downstream (monopsony) market power? Ganapati+McKibbin 3
Competition or bargaining? » Current US policy response: expand competition among supplies Speed up approval process for generic copies of drugs (e.g 2019 Drug Competition Action Plan) » Cross-country observation: many comparable countries "bargain" over generic drug prices Australia, Canada, UK, New Zealand » US Government is a large purchaser of generic drugs No bargaining; no leverage of monopsony power Generics: 90% of prescriptions and 23.2% of expenditure » Usual concerns about bargaining and innovation don’t apply Ganapati+McKibbin 4
Preview Can the US benefit from either bargaining/reducing barriers to entry? 1. How do prices in US markets with few competitors compare with prices in comparable countries? Cross-country comparison of prices by market size Results: monopolies have 3-4 times higher prices in US; Prices converge once there are 6+ competitors 2. Are policies that promote competition or bargaining more likely to be effective at containing prices in these small markets? Structural model with Nash bargaining and a market entry game Counterfactuals (savings to Medicaid): Competition policy: 3%-8% Bargaining policy: 10%-18% Both: 18% Harmonization of entry costs: 6-%16% Ganapati+McKibbin 5
Conceptual framework » Surplus to be split between a buyer and a seller » Relevant welfare: changes in the realized price of off-patent drugs » Final price: p = µ pharmacy × µ PBM × µ wholesaler × µ manufacturer × mc » Our starting point: p = µ value chain ( S ) × mc S : set of drug suppliers Ganapati+McKibbin 6
Data » Prices and number of approved manufacturers » Public data from government regulators and insurers US Medicaid ; UK NHS ; AU PBS ; NZ Pharmac ; BC PharmaCare ; ON Ontario Drug Benefit Robustness: US Medicare Part D & NADAC Wholesale price » Unit of observation - Molecule - Dose - Form (for pills and capsules) FDA definition of a therapeutic substitute for a generic » Off-patent patient administered drugs: drugs >20 years since first approved in the US » Price: Per-pill price, net of all rebates, discounts, and pharmacy dispensing fees, paid by end users and their government Ganapati+McKibbin 7
Example Data - BC - 2016 Molecule Dose Form Approval y US Mfg Medicaid p BC p pyrimethamine 25 tablet 1953 1 605.51 1.43 mebendazole 100 tablet 1996 1 312.69 5.91 penicillamine 250 capsule 1970 1 224.24 3.89 penicillamine 250 tablet 1970 1 90.04 0.68 procarbazine 50 capsule 1985 1 57.06 0.44 morphine sulfate 200 capsule 1987 1 54.05 1.19 methoxsalen 10 capsule 1954 2 49.72 0.65 oxymetholone 50 tablet 1972 1 35.24 1.77 hydromorphone 32 tablet 1926 1 37.20 11.49 ethacrynic acid 25 tablet 1967 2 18.46 0.97 Ganapati+McKibbin 8
Data Table: Medicaid - Molecule-Dose-Form Comparison log ( P US /P Dest ) Start End Adjusted Mean First Mean # Comparison Obs Year Year Mean Std. dev. FDA Approval US Sellers AU 1706 2008 2017 0.505 1.685 1980 4.25 BC 858 2015 2017 0.282 1.783 1983 4.3 NZ 1470 2009 2017 0.332 1.571 1982 4.23 ON 344 2017 2017 0.346 1.578 1984 4.88 UK 1625 2010 2017 0.221 1.839 1981 4.17 Ganapati+McKibbin 9
Key Fact: US prices are high in markets with low competition 3 ln(USA/Foreign Price) Relative to 7+ Mfg 2 1 ln(US/AU Price) ln(US/BC Price) ln(US/NZ Price) ln(US/ON Price) 0 ln(US/UK Price) 1 2 3 4 5-6 US Approved Suppliers Ganapati+McKibbin 10
How important are these small markets? » Our analysis shows Drugs with just 1 supplier: 1% of doses & 10% of Medicaid spending on off-patent drugs. Drugs with <6 suppliers: 25% of doses & 50% of total spending. Ganapati+McKibbin 11
Pricing Model » Key elements: Buyer and a seller bargain over a price Price depends on: competition among sellers bargaining power of buyers » Retail price: p = µ pharmacy × µ PBM × µ wholesaler × µ manufacturer × mc Past literature: uses p ex − manufacturer = µ manufacturer × mc without accounting for either ex-ante or ex-post lump-sum payments. » Our starting point: p = µ value chain × mc What matters for welfare - is final price, not some intermediate price. Ganapati+McKibbin 12
Two-Period Game 1. Generic suppliers choose to enter the market. 1.1 Pay fixed costs (includes regulatory costs, as well as bi-lateral payments to PBMs, pharmacies, doctors, and wholesalers) π f,d ( s ; S ) ≥ F f . π f,d ( s ; S ) : profit of the marginal s th supplier Assumptions: Entry costs independent between markets Unlimited number of potential entrants 2. Suppliers (after all payments), negotiate final price with final buyer π f,d ( s ; S ) = µ f,d ( s ; S ) × c f,d × q f,d ( s ; S ) . 2.1 Will be agnostic on the type of competition 3. Sales are made Most public plans have inelastic demand - shown with exchange rate shocks Ganapati+McKibbin 13
How prices are determined: Monopolist seller and monopolist buyer » Nash surplus between seller s and buyer b NS = ( pq − cq ) w s (¯ p b q − pq ) 1 − w s , p b : Choke price ¯ acquisition price if negotiations break down includes political risk » First order conditions imply: p m = w s ¯ p + (1 − w s ) c. (1) » If w s = 0 , marginal cost pricing: p c = c (2) If w b = 0 , monopolist choke price: p m = ¯ p b (3) Ganapati+McKibbin 14
Generalization » What if there is more than one upstream seller? » How to account for upstream market power? Competition function: θ ( S ) : I + → R ∈ [0 , 1] Maps the number of competitors between monopoly and perfect competition Weights between the Nash solution and perfect competition: p = θ ( S ) p m + (1 − θ ( S )) c (4) » Extensions to Bertrand, Discrete choice, Multiple buyers, Repeated game Details » Intuition: Conditional on the number of entrants, pricing is fully determined. Ganapati+McKibbin 15
Simplification + Empirical Implementation » Assume that the choke price ¯ p is a multiplicative function of the marginal cost: p = γ b c ¯ » Parameterize competition: θ ( S ) = exp ( α log S ) » Define a buyer-specific leverage parameter κ b : κ = [ w s γ b + 1 − w s ] ∈ [1 , ∞ ) Sufficient statistic approach Ganapati+McKibbin 16
Empirical identification: intuition » Relative marginal costs ( c 1 / 2 ): At perfect competition relative price should equal relative marginal cost Extremely common heart and diabetes medication Dozens of entrants -> identifies levels of marginal cost differences » Buyer leverage ( κ 1 , κ 2 ) variation in number of competitors between countries Some drugs have more entrants in different countries Driven by unobservable difference in drug demand (i.e. Australia has relatively higher demand for anti-malaria medication than Canada) » Relationship between price and competitors ( α ) Price variation according to the number of competitors: High relative US prices when there are few entrants Ganapati+McKibbin 17
Role of market size » Excess Profits: Π Under constant marginal costs, how much more operating profit is required to enter a particular country to cover fixed entry costs? » How much more does it cost to enter the US, than other markets? Data for market size in US, UK, AU Ganapati+McKibbin 18
Role of market size » Recover fixed cost differences between two markets: Π excess = Π US ( S US ) − Π Foreign ( S Foreign ) (5) Only done for the marginal generic entrant (as opposed to an incumbent brand) » Bound how many more entrants the US ’could’ support: Π excess ( S ∗ US ) ≥ 0 » Intuition: Revealed preference + backward induction. Ganapati+McKibbin 19
Estimates for α and κ Medicaid Medicare(d) NADAC Molecule-Dose Molecule Molecule Competition α -1.18 -1.43 -1.25 (0.08) (0.14) (0.19) Leverage κ US 5.50 6.38 5.87 (0.38) (0.95) (0.60) Leverage κ AU 1.00 1.02 1.00 (0.00) (0.06) (0.00) Leverage κ BC 1.00 1.00 1.00 (0.00) (0.02) (0.00) Leverage κ NZ 1.73 1.18 1.02 (0.08) (0.15) (0.08) Leverage κ ON 1.09 1.00 1.11 (0.19) (0.09) (0.24) Leverage κ UK 1.66 1.75 1.81 (0.17) (0.26) (0.19) Unbounded estimates: details Ganapati+McKibbin 20
Excess Entry Cost Estimates scenario/est ($M) Medicaid Medicare(d) NADAC Molecule-Dose Molecule Molecule AU 13.68 9.54 5.65 (0.42) (1.20) (1.26) UK 7.94 8.00 7.11 (0.38) (1.12) (1.14) Unbounded estimates: details Ganapati+McKibbin 21
Recommend
More recommend
