DOCUMENT DE TREBALL XREAP2011-02 What if there was a stronger pharmaceutical price competition in Spain? When regulation has a similar effect to collusion Ivan Moreno-Torres (Competition Authority of Catalonia, CRES & GiM-IREA) What if there was a stronger pharmaceutical price competition in Spain? When regulation has a similar effect to collusion Ivan Moreno-Torres∗ May 2011 Abstract This paper examines statins competition in the Spanish pharmaceutical market, where prices are highly regulated, and simulates a situation in which there is unrestricted price competition. A nested logit demand model is estimated with a panel of monthly data for pharmaceuticals prescribed from 1997 to 2005. The simulation indicates that the regulation of prices is similar in its effects to cooperation among producers, since the regulated prices are close to those that would be observed in a scenario of perfect collusion. Freedom to set prices and a regulatory framework with appropriate incentives would result in a general reduction in prices and may make the current veiled competition in the form of discounts to pharmacists become more visible. The decrease in prices would be partially offset by an increase in consumption but the net effect would be an overall decrease in expenditure. The counterfactual set-up would also lead to important changes in the market shares of both manufacturers and active ingredients, and a reversal of generic drugs. Therefore, pro-competitive regulation would be welfareenhancing but would imply winners and losers. Keywords: pharmaceutical industry, statin drugs, competition, regulation JEL classifications: I11, I18, L13, L65 ∗ Competition Authority of Catalonia (ACCO), Generalitat de Catalunya; and research fellow of the Center for Research in Economics and Health (CRES), Universitat Pompeu Fabra, and of the Research Group on Governments and Markets (GiM), Universitat de Barcelona. E-mail: imorenot@gencat.cat Contents 1. Introduction ............................................................................................... 2 2. The Spanish pharmaceutical market......................................................... 4 3. Data ............................................................................................................ 7 4. Empirical framework ................................................................................ 10 Demand................................................................................................................... 12 Supply...................................................................................................................... 18 Simulation................................................................................................................ 21 Welfare variation...................................................................................................... 23 5. Econometric estimation ...........................................................................24 6. Simulation results .....................................................................................27 Prices and expenditure............................................................................................. 27 Active ingredients .................................................................................................... 28 Generics vs. brand-name drugs................................................................................ 30 Firms ....................................................................................................................... 30 Consumer welfare.................................................................................................... 32 Regulation and collusion.......................................................................................... 33 7. Discussion.................................................................................................33 8. Concluding remarks .................................................................................40 Acknowledgments ........................................................................................42 References.....................................................................................................43 Figures and Tables .......................................................................................48 1. Introduction Due to aging of the population and the inclusion of expensive new products, drug expenditure is one of the fastest growing components of health expenditure in most countries. Since the regulation of the market and certain cost containment measures have not been totally successful in reducing prices, knowing more about drug competition could be very useful for pharmaceutical policy makers. Although Spain has allowed the introduction of generic drugs since January 1997, that is, drugs that are bioequivalent to brand-name medicines (i.e., their efficacy and 2 safety are essentially the same) that enter the market when the patents on the original drugs have expired, there is only limited price competition. The pharmaceutical market is highly regulated and prices are driven mainly by modifications and revisions of the regulations. Therefore, the Spanish case allows a study of competition between pharmaceuticals and also simulations of equilibrium prices in situations without price restrictions or with a regulatory framework with appropriate incentives. The main purpose of this paper is to analyse demand for statin drugs and to perform a counterfactual exercise in price competition à la Bertrand. The comparison between the real and the counterfactual situations allows me to evaluate modifications of the strict price regulations in terms of welfare and winners and losers. This new empirical evidence from Spain, which is one of the largest markets for pharmaceuticals in the European Union (EU) and the seventh largest worldwide, could be applied to a great number of countries with similar institutional settings and pharmaceutical market characteristics: heavy price regulation and low penetration of generic drugs. The outline of the paper is as follows. The next section explains the main features of the Spanish pharmaceutical market. The third section describes the dataset. The section after that presents the demand and supply models, and the simulation procedure. The fifth section shows the estimation issues and results. The sixth section presents the simulation results. The section after that discusses the results and the last section offers the concluding remarks. 3 2. The Spanish pharmaceutical market The Spanish NHS is funded from tax revenue and provides health care services to all residents. The health care management system is decentralized and regional authorities control expenditure and organize health service provision. However, the central government regulates pharmaceutical prices. Although the NHS provides health care, there are copayments for prescribed pharmaceuticals. The standard rate of copayment is 40%, but the average copayment is less since prescription drugs for pensioners and some other specific groups, such as the handicapped or people who have suffered occupational accidents, and their dependents, have no charge, and drugs indicated for chronic diseases have a rate of only 10% (with an upper limit). The actual average copayment for prescribed pharmaceuticals is very low and accounts for less than 7% of the total expenditure on ambulatory prescription pharmaceuticals. Another important feature is that there are a great number of different presentations of drugs. In fact, there are three types of prescription drugs in Spain: original brandname drugs (which might be marketed either by the patent holder or by a licensee), copy brand-name drugs and generics. This is due to the fact that, although Spain joined the European Patent Convention in 1986, it did not grant product patent rights until 8 October 1992 due to a transitional period in accordance with Article 167 of that convention. Before this date, there were only process patents. 4 Although generic drugs were introduced into Spain in 1997, the market share for generic medicines was low at the end of the period analysed (14.60% in units and 7.90% in value) (IMS Health, 2006b). The average price of drugs in Spain is low in comparison to that of other EU countries but the average brand prescription price has risen, due mainly to drugs that have been introduced recently at high prices (Costa-Font and Puig-Junoy, 2005). These new high prices may be a strategy to avoid parallel trade, since the low prices for older medicines has led to the Spanish market becoming an important source of parallel trade within the EU. Although Spain was a relatively low-price country with limited generic penetration, in December 1999 a reference pricing system was introduced for off-patent drugs with the same active ingredient. Reference pricing has gradually been extended to a growing list of active chemical ingredients. All versions of off-patent drugs, branded and generics, were included in their respective group of bioequivalent drugs once there was at least one generic version of the respective active ingredient. The reference price was determined endogenously as a function of drug prices in the relevant market: for each group a reference price was calculated as the weighted average selling price of the lowest-priced drug accounting for at least 20% of the market (year on year)1. This system established the maximum price that could be reimbursed by the NHS for any version of the same drug. Whenever the price of a If the difference between this price and the highest price for the group was less than 10%, the reference price was the result of applying a 10% reduction to the highest price. If the difference between the calculated price and the highest priced product was more than 50%, the reference price was exactly 50% of the highest priced product. Whatever the situation, the reference price was never lower than the price of the cheapest generic (López-Casasnovas and Puig-Junoy, 2000). 1 5 prescribed drug was higher than the reference price, patients could opt for the prescribed drug by paying the difference between its price and the reference price. However, since January 2004 the reference price has been calculated as the average of the three lowest costs per day of treatment, for each form of administration of an active ingredient, according to its defined daily dose2 (DDD). With this system, if prescriptions specify drugs priced higher than the reference price, pharmacists are obliged to substitute them for the cheapest generic version. However, if prescriptions specify drugs whose price is equal to or lower than the reference price, pharmacists are not obliged to substitute them. When the prescription has been written using the name of the active ingredient, the pharmacist has to dispense the lowest-priced generic drug. In this way, reference pricing has become a system for establishing the maximum reimbursement price that a drug may have without being excluded from the list of publicly financed drugs, that is, a kind of price capping system. Moreover, the maximum ex-factory price for all drugs (branded and generics) is set during the process of obtaining market approval, and usually the introduction price remains the maximum price for most of the life of the product (Borrell, 2003). The Ministry of Health and Consumer Affairs is responsible for negotiation with firms. The funding conditions setting and wholesalers’ and retailers’ mark-ups are also regulated. 2 A defined daily dose is the average dose per day in adults for a drug when it is used to treat its main indication. 6 The government uses a peculiar form of cost-based price regulation for branded drugs in which manufacturing, marketing and research costs, as well as an industrial profit on invested capital, are allocated to new drugs. However, this is rarely the final price since the legislation allows other factors to be considered such as the price of the same product in other European countries, the price of drugs that can be considered substitutes or the therapeutic innovation of the medicine. In fact, the legal criterion is that the price has to reflect the therapeutic value of the drug as well as the cost of comparable treatments, the price of the same drug in other countries, and some other political issues such as the contribution to the national economy (Antoñanzas et al., 2007). Finally, it is important to emphasize that, in Spain, public expenditure on prescription medicines represented slightly more than 22% of public health expenditure in 2005, and that between 1997 and 2005, the public expenditure on pharmaceuticals increased by almost 117%. 3. Data The dataset consists of monthly consumption records of prescription statins from 1997 to 2005. The information was provided by the Directorate-General of Pharmacy and Health Products at the Spanish Ministry of Health and Consumer Affairs, and is complemented with data from the Nomenclator Digitalis of the NHS Health Information Institute (Ministerio de Sanidad y Consumo, 2005) and from the Base de Datos del Conocimiento Sanitario 2005 - BOT PLUS (Consejo General de Colegios Oficiales de Farmacéuticos, 2005). 7 The analysis is focused on outpatient consumption of oral prescription drugs containing only one active chemical ingredient. As Table 1 shows, the sample includes 19 markets for six active ingredients from the group of statins (HMG CoA reductase inhibitors). The active ingredients included are all those commercialized in Spain during the period analysed: atorvastatin, cerivastatin, fluvastatin, lovastatin, pravastatin and simvastatin. However, cerivastatin (marketed in Spain as Lipobay®, Liposterol®, Vaslip®, and Zenas Micro®) was withdrawn from the market in 2001, due to international reports of safety issues. These drugs are lipid modifying agents that are used to lower high cholesterol levels in people with or at risk of cardiovascular disease. In fact, they are the most potent cholesterol-lowering agents available. The reduction in total cholesterol levels that these drugs are responsible for, and especially in levels of low-density lipoprotein (LDL) cholesterol (commonly known as "bad cholesterol"), implies an important decrease in the number of cardiac events, such as heart attack or sudden cardiac death, and a considerable reduction of the risk of stroke. For this reason, statins are widely prescribed and among the most sold drugs worldwide. Lovastatin, simvastatin, pravastatin and fluvastatin entered the market before January 1997. Lovastatin was the first statin marketed in Spain; in 1990. In 1991, simvastatin and pravastatin were commercialized. Fluvastatin entered the market in 1996. The first sales of atorvastatin and cerivastatin appear in the dataset in November 1997 and August 1998, respectively. 8 Table 1 also shows the specific indications of each active ingredient3, the date of entry of the first generic (in those markets with generic entry before the end of 2005), and the date of the implementation of the reference pricing system (for those markets in which this system of reimbursement was applied). This dataset is not a sample but the whole outpatient market for these drugs: all statin drugs sold in Spain and financed (at least partially) by the NHS. The panel is unbalanced since atorvastatin and cerivastatin entered in the market after January 1997 and different presentations entered the market at different times. There are, however, observations for at least 56 months for cerivastatin, 98 for atorvastatin and 108 for the remaining active ingredients. There are a total of 10,981 observations for 64 producers. The different active ingredients have several strengths (quantity of active ingredient per unit) and are sold in different package sizes (number of units); therefore, all the quantities sold are converted into common units. For each presentation I calculate the total number of milligrams of the active ingredient and transform this into patient days using the DDD; that is, I calculate the total number of DDDs per package4. Prices are calculated from the dataset by dividing volume of sales in euros (€) by the number of DDDs sold: the price per DDD for each product. The indications for the prescription of statins have broadened over the years, to include for instance the preventative effects of statin use in specific risk groups, such as diabetics (Collins et al., 2003). 4 I use the DDD of each active ingredient established by the World Health Organization (WHO) Collaborating Centre for Drug Statistics Methodology. The use of this international standard dose enables standardisation and comparison drug quantities across therapeutic groups, active ingredients and presentations. 3 9 Finally, I calculate the market size as the population who have high cholesterol and therefore, could potentially receive a prescription for a statin. To calculate this potential market I use the number of individuals reported to have been diagnosed with high cholesterol in the 2003 Spanish National Health Survey (Encuesta Nacional de Salud). I assume the percentage of population with high cholesterol to be constant over the years analysed and I multiply this percentage by the Spanish population and the number of days in each month. In this way I obtain the number of DDDs potentially consumed. 4. Empirical framework Two of the most important drivers of pharmaceutical competition are the introduction of new active ingredients for similar indications and the introduction of generic drugs. As mentioned, generics were introduced into Spain in 1997. These products contain the same active chemical ingredient as the original product and have proven bioequivalence5, to the satisfaction of health authorities. However, the producer and some characteristics such as colour, shape, inactive ingredients or packaging may be different. Therefore, generics may be considered substitutes for the brand-name drugs but not perfect substitutes; in other words, there is some degree of product differentiation. The WHO defines two pharmaceutical products as bioequivalent if they are pharmaceutically equivalent or pharmaceutical alternatives, and they display comparable bioavailability, when studied under similar experimental conditions. Bioequivalence is considered proven if the bioavailabilities, in terms of peak and total exposure after administration of the same molar dose under the same conditions, are similar to such a degree that the effects of the studied products can be expected to be essentially the same. 5 10 The expected result of generics entering the market is an increase in the level of competition in the market and a reduction in the prices and market shares of brandname products. It is also expected that when the number of manufacturers producing generics in the market is considerable, prices will tend towards the marginal cost of production. However, for brand-name products, the literature in the US shows some evidence of price increases after the entry of generics (for instance Grabowski and Vernon, 1992; or Frank and Salkever, 1997): this has been called the “Generic Paradox” (Scherer, 1993). Product differentiation is greater if not only the producer, colour, shape, inactive ingredients or packaging are different, but so too is the active ingredient. Indeed, each statin has its own characteristics and entered the market at a different time. For instance, statins may be classified into fermentation-derived and synthetic. The first group includes lovastatin, pravastatin, and simvastatin; the second atorvastatin, cerivastatin, and fluvastatin. The LDL-lowering potency also varies from one active ingredient to another. Cerivastatin was the most potent, followed by atorvastatin, simvastatin, lovastatin, pravastatin, and fluvastatin. A comparison of the efficacy of atorvastatin, simvastatin, pravastatin, lovastatin, and fluvastatin at reducing LDL and total cholesterol in patients with hypercholesterolemia, found that atorvastatin was the most effective without increasing adverse effects (Shepherd et al., 2003). Finally, the 11 statins analysed are generally well-tolerated and show similar levels of adverse effect, although the newer statins have a higher ratio of efficacy to adverse effect6. Therefore, the statin group is formed of branded and generic medications that compete with other close but not perfect substitute medications. Thus, defining the relevant market is not easy and although it is an imperfect approach, some authors (for instance, Aronsson, Bergman and Rudholm, 2001; Dalen, Strøm and Haabeth, 2006; and Moreno-Torres, Puig-Junoy and Borrell, 2009) take the therapeutic active ingredient level according to the Anatomical Therapeutic Chemical (ATC) classification as the relevant market. I follow this approach to define markets but, in order to tackle the non-linearity of the prices of differently sized presentations and to take advantage of variation between markets and products to identify parameters, I use the product (active ingredient with a specific dosage and size from each manufacturer) as the element of analysis. Demand In the pharmaceutical industry the institutional setting is very important and the dispensing process is quite complex; the physician, the pharmacist, the third-party payer and the patient each play a role. I assume that the physician and the pharmacist are perfect agents for the patient and that their choice, together with that of the patient, maximizes utility for the patient. The risk of myopathy is lowest with pravastatin and fluvastatin probably because they are more hydrophilic and as a result have less muscle penetration. Lovastatin induces the expression of gene atrogin-1, which is believed to be responsible in promoting muscle fibre damage (Hanai et al., 2007). 6 12 The dispensing process follows a multi-level market structure, as shown in Figure 1. The first choice is between dispensing a statin or prescribing an alternative treatment, such as exercise, a change of diet, homeopathic products or even a drug from another related therapeutic subgroup such as fibrates, bile acid sequestrants or nicotinic acid and its derivatives. In fact, prescription guidelines usually require that the patient has tried a cholesterol-lowering diet before starting to use statins. If the statin subgroup is chosen, the next step is to decide between the active ingredients based on the medical record of the patient and possible intolerance to some active ingredients. The third decision is to choose a specific presentation from all the possibilities for the active ingredient, in other words, to choose the strength of the product and the package size, given the difference between the actual and targeted cholesterol level, as well as the characteristics of the patient. The next stage is the choice between branded and generic drugs. Hellerstein (1998) found that physicians play an important role in determining whether patients receive brand-name or generic pharmaceuticals and that some are more likely to prescribe generics while others are more likely to prescribe branded products. Coscelli (2000) found that in addition to the physician, the patient’s characteristics also affect the prescription decision. When the physician prescribes a brand-name or generic product, there is also a choice between different manufacturers. In fact, as well as the physician and the patient, the pharmacist may participate in the choice of the drug, for instance choosing which product containing the given active ingredient to dispense. The 13 choice at this stage is mainly based on which products are available at the pharmacy. Finally, the third-party payer may affect the decision by implementing obligatory substitution rules and requirements limiting reimbursements, such as the reference pricing system. To analyse this market I use a structural discrete-choice model of product differentiation. In this model utility for consumers depends on product characteristics and individual tastes, firms are modelled as price-setting oligopolists and endogenous market outcomes are derived from an assumption of a Nash equilibrium in prices. This kind of random utility model has been applied to a great number of products such as ready-to-eat cereals (Nevo, 2000a and 2001), yogurts (Di Giacomo, 2008), movies theatres (Davis, 2006) and vehicles (Berry, Levinsohn and Pakes, 1995; Verboven, 1996; or Petrin, 2002). They have also been used to analyse demand for pharmaceuticals by several authors such as Stern (1996), Cleanthous (2004), Dalen, Strøm and Haabeth (2006), Iizuka (2007), Yu and Gupta (2008), Kaiser, Mendez and Rønde (2010) and Coronado (2010). As my interest is not only in competition between drugs containing the same active ingredient, but also between drugs with different active ingredients, I use a nested logit model based on Berry (1994)7. In contrast to the simple multinomial logit model, the nested logit model allows consumer tastes to be correlated across products. Indeed, there is correlation between the idiosyncratic shocks between products of the same segment of the market. This counters the “independence from An alternative approach is the multistage budgeting model applied by Ellison et al. (1997) to analyse the demand for cephalosporins. 7 14 irrelevant alternatives” (IIA) property and allows for reasonable substitution patterns. In comparison to the random coefficients model8, the nested logit model may be sensitive to the specification of the nest structure because the researcher chooses the options that are potentially close a priori9. However, in the case of pharmaceuticals this problem does not jeopardize the nested logit model since it is possible to use the ATC classification to build the structure of the nests10. It seems reasonable to assume that, within statins, products with the same active ingredient are closer substitutes than products with different active ingredients. Additionally, with the random coefficients model it is necessary that prices or other product characteristics vary across markets (Nevo, 2000b) and in the dataset there are not such variations. Drugs are grouped into exhaustive and mutually exclusive active ingredient markets. In each of these market, for instance “lovastatin”, there is a set of drugs characterized by their strengths, package sizes and manufacturers denoted j = 1, …, J. Each set may include brand-name and generic drugs. It is assumed that different presentations with the same active ingredient are highly substitutable and may be The nested logit model can be interpreted as a special case of the random coefficients model with random coefficients only on segment–specific dummy variables (Berry, 1994). The consumer preference for the segment is the only relevant consumer characteristic and it interacts with only one product characteristic: the segment to which the product belongs. 9 For those cases in which the order of the nests is important, Bresnahan, Stern and Trajtenberg (1997) developed the principles-of-differentiation general extreme-value model. 10 The therapeutic groups are formed on the basis of the 4th level of the ATC code that approximates the chemical, therapeutic or pharmacological group (e.g., C10AA for statins or HMG CoA reductase inhibitors) and the pharmaceutical presentations within a group can be considered close substitutes since they have the same active chemical ingredient (5th level of the ATC code, e.g., C10AA01 for simvastatin). 8 15 interchanged; for instance, one daily pill of simvastatin 20 mg with either two pills of simvastatin 10 mg or half a pill of simvastatin 40 mg. Finally, the outside good represents the alternative: not consuming a statin, and is assumed to be the only member of its own group11. The physician, jointly with the pharmacist and the patient, chooses a presentation of the drug that maximizes utility for the patient, even though the final consumption may be affected by the reimbursement and substitution rules and the availability of the product at the drugstore. The aggregate demand for each product is obtained by summing over the individual choices. The indirect utility function of consumer i for consuming the drug j at the period t is: u ijt = δ jt + ∑ d jg ζ ig + (1 − σ )ε ijt g [ ] (1) Where δ jt is the mean utility level of product j, which is the same for all consumers. Thus, individual heterogeneity enters the model through the non-deterministic part of the utility, ζ ig + (1 − σ )ε ijt . Where (1 − σ )ε ijt is the idiosyncratic taste for drug j and ζ ig is the individual-specific random term, which is interacted with the dummy variables for each of the active ingredient markets. In fact, ε ijt is an identically and independently distributed extreme value random variable that represents the distribution of consumer preferences about the mean utility, d jg is a dummy variable In fact, the outside good also includes the consumption of statins by those individuals with private health insurance. 11 16 that is equal to one for drugs with a specific active ingredient, g, and zero otherwise and ζ ig is common to all products in market g and has a distribution function that depends on the parameter σ, with 0 ≤ σ < 1. As Cardell (1997) proved, an additional property of this model is that if ε ijt is an extreme value random variable, [ ζ ig + (1 − σ )ε ijt ] is also an extreme value random variable. The parameter σ measures the within-nest correlation of utility levels and allows substitution between products with the same active ingredient to be included. If σ approaches 0, the within-group correlation of utility levels is low, and the model tends toward the multinomial logit (Besanko, Gupta and Jain, 1998). When σ tends toward 1, the within-nest correlation of utilities approaches 1. The mean utility, δ j , is equal to: δ jt = κ + γ g + Xβ + αp jt + ω m + ξ jt (2) Where the κ is a constant term, γ g is an active ingredient fixed effect, X is a matrix of product characteristics which includes the time of each product in the market and a dummy variable that is equal to 1 if the product is a generic, α is the price coefficient and β is a vector of taste parameters to be estimated. p jt is the price per DDD of drug j at time t. I also introduce firm-specific fixed effects, ω m , that is, a variable that is equal to one for a specific firm across drugs and markets and to zero otherwise. ξ j expresses product characteristics that are observable to 17 the physicians, pharmacists or patients, but not to the researcher, such as quality, reputation or promotional effort. One characteristic of this demand model is that wealth effects are not taken into consideration. This would be a problem if the product analysed was a good whose price sensitivity depended on level of income, such as a durable good, but this does not seem to be the case for statin drugs. After the normalization of the utility of the outside good to 0, δ 0 = 0 , and following the derivation in Berry (1994), the econometric specification is: ln( s jt ) − ln( s0t ) = κ + γ g + Xβ + αp jt + σ ln( s j / gt ) + ω m + υ jt (3) Where s jt and s0t are the market shares of drug j and the outside good, respectively, σ is the parameter that measures the level of correlation between products with the same active ingredient, ln(s j / gt ) is the log of the share of product j in the group of products with active ingredient g, and υ jt is an error term. Supply In the pharmaceutical industry, firms usually produce more than a single drug. I assume that firms act as multi-product Bertrand profit maximizers and set their prices taking the prices chosen by their competitors as given. Following Nevo (2000a and 2001), each firm produces a subset Γ f of the J products. So, for each period, the profits of the firm f are: Π ft = ∑(p j∈Γ ft jt − mc jt ) Ms jt ( p ) − C ft (4) 18 Where s jt ( p) is the market share of product j, which depends on the prices of all products, and M is the size of the market, which includes the outside good and implicitly defines its size. As highlighted by Nevo (2000a), this definition allows the market size to be kept fixed while still allowing an increase in the number of the products on the market, since it results in a decrease in the share of the outside good12. C ft is the fixed cost of production and mc jt is the marginal operating cost, which is assumed not to depend on the quantity produced. A unique pure-strategy Bertrand-Nash equilibrium with strictly positive prices is assumed. Thus, the price p jt of product j produced by firm f at period t must satisfy the following first-order condition: s jt ( p ) + ∑(p r∈Γ ft rt − mcrt ) ∂srt ( p ) =0 ∂p jt (5) These first-order condition equations involve price-costs margins for each drug. Bertrand competition in a context of differentiated products, which is the case in the pharmaceutical industry, is different from a situation of homogenous products. With differentiated products, a firm does not generally lose all of its demand by pricing slightly above the competitors’ prices, nor does it steal all rival firms’ demand by pricing below their competitors’ prices. Thus, in equilibrium, firms may set different prices that exceed the marginal cost and earn positive profits (Baye and If the potential market size (which was defined in section 3 as the amount of DDDs that would be consumed if all potential patients took a statin) was reduced, for instance, by considering a lower proportion of people with high cholesterol level as potential consumers, it would imply higher elasticities of substitution. Since I assume a relatively large market size for the outside good, it will generate conservative estimates of substitution. 12 19 Kovenock, 2008). If products are considered good substitutes, there is more competition and prices should tend towards the marginal cost: with less differentiation, there is a reduction of markups and prices. For instance, the differentiation between brand-name and generic drugs may explain the aforementioned “Generic Paradox” found in the US by some authors. The markups can be solved for explicitly by defining the following J x J matrix:  ∂srt ( p) , if − Ω jrt ( p) =  ∂p jt 0, otherwise  ∃f : {r , j} ⊂ Γ ft (6) Where the own and cross price elasticities of this one-level nested logit model have the following form: α p jt (1 − σs j / gt − (1 − σ ) s jt ) 1−σ α 1−σ η jjt = η jrt = prt (1 − σsr / gt − (1 − σ ) srt ) (7) η jkt = −αpkt s kt Where j and r are different products with the same active ingredient and k has a different one. The first-order conditions in (5) can be written in vector notation as: s ( p ) − Ω( p − mc) = 0 (8) 20 Simulation If the marginal costs are not observed, they can be computed from the estimates of the demand system and (8). However, I follow the opposite approach and obtain the vector of prices for each month of 2005 from the inversion of Ω , the market shares and a proxy of the marginal costs. Once the own and cross price elasticities are computed and Ω is inverted, I calculate the prices using the following expression: p = mc * + Ω −1 s ( p ) (9) As a proxy for the marginal operating costs, I use the lowest price per DDD observed for the 19 markets in August 2010, which is €0.07 for the presentation of simvastatin with a dose of 40 mg and 28 tablets. It seems a reasonable marginal cost since the prices tend to the marginal cost when there are a considerable number of competitors and usually, in regulated markets, some years after the introduction of the product. In any case, this price is assumed to be a maximum boundary for the marginal cost. Additionally, I compute prices with a marginal cost equal to zero as a minimum boundary. In this way, I assume that marginal costs are constant across firms and periods. The assumption of constant marginal cost of production is common in the literature that analyses the pharmaceutical industry. It is also common to assume zero marginal cost, since previous observers have claimed that the marginal cost is extremely low, even that the level of marginal cost is negligible compared to price (Stern, 1996). Therefore, the marginal operating cost per DDD will be €0.07 or zero. 21 For practical purposes I assume that in each period firms set prices taking into account their own and competitors’ market shares in the previous month. After calculating the simulated prices, I compute the corresponding market share of each product for that period with the parameters obtained from the estimation of the demand model. With these market shares, I compute the prices of the following month in a recursive way and with them the market shares for that period13. As an alternative scenario, prices in 2005 are computed only for off-patent drugs14. Taking the regulated prices of drugs under patent protection as given, I simulate the prices for off-patent products in each period using the previous procedure. Then, I recalculate the market shares of all the products, including those with patent protection. Finally, I simulate a situation of full collusion among all the firms present in the market. Thus, in these scenarios each producer considers in its first-order condition of profit maximization not only all the products it produces but also its competitors’ products. I simulate the prices resulting from this strategy of joint profitmaximization of all the products, which corresponds to monopoly or perfect price collusion, by defining the ownership matrix appropriately and computing the corresponding markups (Nevo, 2001). I also assume the same number of entries of firms throughout 2005, which is a small number, and that they achieve the same market share in the first month as in the real data. 13 14 The inclusion of all the products in the first counterfactual scenario does not mean that active ingredients whose patents have not expired lose their protection, since the entry of new competitors, for instance generics, is not simulated. However, these products are under competitive pressure from other active ingredients in the same therapeutic group and, as will be seen below, their producers are in a better position if they are able to set prices without restrictions when off-patent products can do so. 22 Welfare variation In order to evaluate the effects of introducing actual price competition, it is necessary to calculate variations in welfare. Welfare is a monetary measure of the utilities for the patients of the set of products available, including the outside good, minus the price paid by the consumers (the copayment of the patients plus the cost for the NHS). Some previous work has analysed effects on welfare of the introduction of new products (Petrin, 2002; Di Giacomo, 2008). I only compute welfare variations as a result of price changes for the same set of products, for which it is necessary that there are no welfare variations due to variety effects (Hausman and Leonard, 2002). Following Small and Rosen (1981), the surplus per consumer in a discrete-choice model may be calculated by integrating over the market share function. Thus, the compensating variation measure of a change in prices for a representative consumer is equal to: 1 δ t1 0 t Wt = α ∫δ s jt (δ r )dδ jt (10) Where δ0 is the vector of mean utilities calculated with the real prices and δ1 is the same vector computed with simulated prices of each of the counterfactual scenarios. After integrating the market share formula for the one-level nested logit model (Berry, 1994), the following expression is obtained: 23   δr 1    ∑ exp jt Wt = ln ∑ 1 − σ α  g  j∈g        1−σ   + 1   δ t0 δ t1 (11) Where g is the active ingredient subgroup. The total variation in welfare is given by multiplying this compensating variation for a representative consumer by the potential market size. I use the parameter estimates of the demand model to calculate expression (11). Finally, since the entry costs are sunk costs and fixed costs are the same under any scenario, it is possible to calculate the welfare variation of firms during 2005 as the change in variable profits, i.e. the quantity of DDDs sold multiplied by the difference between the price per DDD and the assumed marginal cost per DDD. 5. Econometric estimation The quality of the product as well as promotional activities and marketing expenses are observable to the consumer, the physician and the pharmacist, but not to the researcher. These unobserved factors are correlated with the drug price and with the drug’s share within the active ingredient and cause endogeneity. To partially overcome this problem, I include firm-specific dummy variables in the econometric specification. The role of these fixed effects is to control for time invariant factors, which are usually common to all a firm’s products, that is, to capture the manufacturers’ mean quality and mean marketing effort, leaving the time-specific and product-specific deviations as part of the error term. In fact, the 24 market shares of competing drugs are driven by physicians’ and pharmacists’ choices of producer within the market and firms build a reputation and develop idiosyncratic skills in launching and delivering drugs. So, the remaining potential endogeneity arises from factors that change over time or variation between a manufacturer’s products. To deal with this remaining inconsistency in the estimation, I use instrumental variables. Following the empirical literature on industrial organization (Berry, Levinsohn and Pakes, 1995; Bresnahan, Stern, and Trajtenberg, 1997; and Petrin, 2002) and similar papers dealing with the pharmaceutical industry (for instance Iizuka, 2007; or Stern, 1996), as possible instruments I considered those variables that capture how crowded product space is and the ownership structure at different levels of the market. From the instruments available I chose the set with strongest correlation to the endogenous variables that did not reject the null hypothesis of the Sargan-Hansen test of exogeneity (Sargan, 1958; and Hansen, 1982). These instruments are: total number of products in the active ingredient market; total time in the market of all a firm’s products; and total time in the market of all the competitors’ products. Additionally, I include the order of entry of each product into the active ingredient market as an instrument. In Spain, as Reiffen and Ward (2005) indicated for the US, the timing of entries into the market is largely out of the control of firms; they do not know the date of approval with certainty, or even if they will obtain approval, 25 neither do they know when, or how many, other applications for that market will be approved. Thus, because of the length and uncertainty of the approval process, order of entry may reasonably be considered as unrelated to firm-specific characteristics. I use the (two-stage) Generalized Method of Moments (GMM), which is the method most commonly used in the industrial organization literature following Berry (1994), since its estimates are unbiased and consistent. Table 2 contains the definitions and descriptive statistics of the variables that are used in the estimation. Table 3 shows the GMM estimation of the demand model. The within-activeingredient market share is significant and its coefficient is 0.46. This result validates the use of the nested logit model as opposed to a simple multinomial logit model, since the nested model is only consistent with the random utility maximization when this parameter is significant and between 0 and 1. There is negative price elasticity since the price per DDD parameter is significant and negative. A 1% increase in price reduces the relative statin drug share, that is, the weight of the product share over the outside good share, by 3.46%. Time in the market has a positive coefficient and is significant. It would seem that products that have been on the market longer have more sales than new products. The generic dummy variable is significant at the 5% level and has a positive parameter. Thus, being a generic product also seems to have a positive effect on the quantity of DDDs sold. 26 As expected, the dummy variable for the general withdrawal of cerivastatin until its complete disappearance from the market has a significant large negative coefficient. The remaining active ingredients, except for lovastatin, have significant dummy coefficients in relation to the comparison active ingredient, simvastatin. In fact, atorvastatin and fluvastatin have positive and larger values. 6. Simulation results Once the own and cross price elasticities given by (8) are computed and Ω is inverted, I calculate the price of each product for each month of 2005 using (9). As proxies for the marginal cost I use the lowest price per DDD observed for the 19 markets in August 2010 and zero. After calculating the simulated prices I compute the corresponding market shares for each product for that month. The 2005 prices are also computed only for off-patent drugs, in order to simulate a scenario in which there is price competition only among products that are not protected by patents. With these counterfactual prices and market shares, which are obtained from a competition à la Bertrand, interest lies in the comparison with the real ones, which are the result of a competition restricted by price among other regulations. Prices and expenditure As Table 4 shows, there is a global decline in unrestricted prices15. The weighted average real price per DDD is €0.44 for the whole statin market, while the mean simulated price goes from €0.18 (when the marginal cost is assumed to be zero) to In order to homogenize comparisons with the real prices and quantities, the real quantities of DDDs are obtained from the predictions of the estimated demand model with the real prices and the real average prices per DDD are weighted by these predicted quantities. 15 27 €0.25 (for a marginal cost of 0.07). If there is price competition only between offpatent drugs, the average prices are €0.21 in the first case and €0.27 in the second. Therefore, the price reduction ranges from 39% to 59% depending on the scenario. As Table 7 shows, due to the negative price elasticity of demand, there is a slight increase of about 4% in the total quantity of DDDs consumed under any scenario. However, the price decline is enough to compensate this increase in consumption and, as is shown in Table 5, there is a considerable decrease in expenditure, especially in those cases where there is no marginal cost. The savings for the NHS and patients ranges from 37% to 57%, depending mainly on the marginal cost assumption. Active ingredients Even though there are considerable real price differences between active ingredients, once there is unrestricted price competition, prices are quite similar. The lowest real price is €0.36 for atorvastatin and the highest is €1.00 for pravastatin. When there is competition between all the products and the marginal cost is assumed to be €0.07, the price per DDD is €0.24 for fluvastatins and lovastatins, €0.25 for simvastatins, and €0.26 for atorvastatins and pravastatins. If the marginal cost is assumed to be zero, prices vary between €0.17 and €0.19. When the competition is restricted to off-patent products, these drugs have almost the same mean prices (in fact they are identical when expressed to two decimals) while the products under patent protection (atorvastatin and fluvastatin) have the 28 regulated prices. For that reason, the decline in prices is slightly lower for the whole statin market. Tables 5 and 6 show the distribution of total sales and market shares among active ingredients for each scenario, respectively, and Tables 7 and 8 show the same information in terms of quantity of DDDs. The data in these four tables are represented in Figures 2 to 516. As a consequence of price variations, there are important changes in the total amount of DDDs and market shares of active ingredients. On the one hand, in all the scenarios there is a large increase in consumption of pravastatins, especially when the competition is restricted to off-patent drugs. On the other hand, there are large declines in sales of atorvastatins and simvastatins. The consumption of fluvastatins and lovastatins are relatively stable and fluctuate depending on the assumptions of competition and marginal cost. When the competition is restricted to off-patent drugs, the decline in sales of products under protection is greater. However, as they have relatively high regulated prices, the revenues they generate are less affected than the quantities sold. Due to space limits, variations in the market shares of different presentation (combinations of dose, quantity and active ingredient) for the same active ingredient are not reported. Active ingredients with considerable variations, in terms of quantities as much as revenues, are atorvastatin, fluvastatin and pravastatin. For the first two active ingredients, not only are there variations in comparison with the real scenario, but also between the situation in which price competition is restricted to off-patent drugs and the one in which it is allowed among all products. These results are available upon request. 16 29 Generics vs. brand-name drugs Taking into account that in 2005 there were no atorvastatin and fluvastatin generics because of patent protection, with the real prices, the generics market share for the whole statin market was 13% in value and 12% in quantity. However, with the introduction of price competition there is a clear decline in these market shares. When there is price competition between all the products, the generics market share is 5% in value and 6% in DDDs, and when there is competition only for off-patent products, the market share is 6% in revenue and between 7% and 8% in quantity (Tables 9 and 10). This global decline in generic products is represented in Figures 6 and 7. Among active ingredients, whereas the reduction in the generics market share is not very high for lovastatin and simvastatin (one percent for lovastatin and three percent for simvastatin), it is very marked for pravastatin. In value, the share goes from 34% to 5-6%, and in DDDs, from 41% to 6-7%. Firms As Figures 8 to 12 show, in general terms, the distribution among firms of DDDs sold and revenue are slightly more unequal in any of the hypothetical scenarios than they are with the real prices. In terms of quantity of DDDs, as mentioned above, there is a total increase. However, of the 61 firms active during 2005, only 11 increase their sales when there is competition among all the products, independently of the assumed marginal cost; this becomes 17 or 18 when there is only price 30 competition for off-patent drugs and the marginal cost is €0.07 and zero, respectively. In terms of revenue, in spite of the growth of demand and due to the general decline in prices, there is an overall total reduction. These smaller revenues are more concentrated in the hands of some manufacturers, especially for lovastatins and pravastatins (Figures 13 to 17). In fact, five of the 61 firms active during 2005 register an increase in their revenues when there is competition for all the products and the marginal cost is €0.07; and three firms do so when there is no marginal cost. When competition is restricted to off-patent products, four producers also experience increased revenue, independently of the marginal cost. The reduction of aggregate revenue that is reported in Table 5 is from €44,381,547.88, when there is only price competition among off-patent drugs and the assumed marginal cost is €0.07, to €68,609,032.75, when there is competition for all the products and the marginal cost is equal to zero. The Table 11 shows the change in variable profits among active ingredients and overall. The reduction in variable profits is from €31,465,707.95, in the case of price competition only among products not protected by patents and a marginal cost of €0.07, to €38,196,118.95, when there is competition among all the products and the marginal cost is zero. In percentage terms, the variable profits decline by between 36% and 57%. 31 There is a clear decline in all the active ingredients except for pravastatin, which sees an increase in the number of DDDs sold. In fact, pravastatin experiences a large increase in variable profits in the two hypothetical scenarios in which the marginal cost is assumed to be €0.07. In contrast, the reduction in simvastatin variable profits is very large under all the scenarios and especially when the marginal cost is equal to zero. For atorvastatin, the decline is also large when the marginal cost is assumed to be zero as it is when the marginal cost is equal to €0.07 and all the products experience price competition. In the case of lovastatins and fluvastatins, there are considerable but smaller reductions of variable profits. Consumer welfare Table 12 shows the computed changes in the welfare of consumers, understood as the NHS (tax-payers) plus the patients, under each scenario. The price reductions under all the counterfactual scenarios lead to an increase in surplus. This increase is very great when the marginal cost is equal to zero and especially when all the products compete with each other, i.e., in cases in which the decline in prices is great. The decrease in variable profits is also greater under such scenarios. The net effect of the increase in consumers’ welfare and the decline in firms’ profits is clearly positive in any situation. The net welfare increase ranges from a maximum of €717,343,568.78, when the marginal cost is zero and all the drugs compete, to €538,482,214.06, when the marginal cost is €0.07 and only off-patent products compete. 32 Regulation and collusion As Table 13 shows, average real prices, which are the regulated prices, are very close to those obtained from the simulation of a situation of full collusion among all the firms in the market (joint profit maximization). For the whole market, the weighted mean real price per DDD is €0.44, whereas the weighted mean price resulting from perfect collusion ranges from €0.38, when there is competition only between off-patent drugs and the marginal cost is assumed to be 0, to €0.48, in the scenario where there is competition among all the products and the marginal cost is €0.07. 7. Discussion As the estimation results show, the within-active-ingredient market share has a significant coefficient of 0.46. Since a value of 1 would indicate a situation of perfect correlation among utilities and a value of 0 no degree of possible substitutions at all, this value shows a considerable degree of differentiation within the active ingredients. The price has a negative effect on market share, which means that, even though the rate of copayment is quite low in Spain, there is negative price elasticity. The time in the market has a positive effect on market share. This result is not surprising since in the pharmaceutical market, experience and product reputation are appreciated by patients and healthcare professionals, and there is evidence of an advantage for the first entrants (Yu and Gupta, 2008). 33 A surprising result is that being a generic drug has a positive effect on market share. It is expected that physicians and patients may prefer brand-name medicines, which usually have a better reputation. The coefficient of this dummy variable may capture the efforts of national and regional health authorities in the promotion of the prescription and consumption of generics. One limitation of the estimation of the demand model is that the dataset contains no characteristics of the patients or other agents and thus demand depends only on prices and other attributes of the products. Another pitfall of the dataset is that price rebates are not reflected in it. The prices used in the estimation, and indeed in the real situation in the simulation, are the prices that the NHS actually paid (and the patient when there was a copayment). However, there is some evidence of price discounts in Spain (Borrell and Merino, 2007; Puig-Junoy, 2009), especially for generic medicines, as occurs in other regulated pharmaceutical markets (Kanavos and Taylor, 2007). These price rebates to pharmacists may distort the estimated price elasticity and the subsequent simulations. As expected, with the simulation of Bertrand price competition, in comparison to the situation of highly regulated prices, there is a notable reduction in prices and, although the price elasticity is negative, the increase in consumption is not enough to compensate for the lower prices. So there is a considerable reduction in expenditure for the whole market. 34 Among active ingredients there are important variations in market share. One possible explanation is that under the scenarios of unrestricted price competition there are fewer differences between active ingredient prices and the changes in sales may correspond to other characteristics. Moreover, as the simulations show, when price competition is restricted to offpatent drugs, the decline in sales of products under protection is greater and the overall saving is lower. For this reason, a pro-competitive reform should allow the freedom to set prices for both kinds of active ingredients: with and without patent protection. This would not mean that the former lose their protection, since there is no advance entry for generics. However, it is harder for those active ingredients in a given therapeutic group that enter the market later to recover their R+D expenses since they experience competitive pressure from older active ingredients in markets which may include lower priced generics. On the other hand, their contribution to health improvements is usually also relatively small. In relation to the reduction of the market share of generics, brand-name producers react to the introduction of generics by reducing prices and the generic products lose their comparative advantage, which is a lower price. Thus, this effect is especially great in those scenarios in which all products compete. The reduction in prices shown in the previous section may be understood as the capacity to reduce markups from the current regulated prices. This means that if rebates are a common strategy of statin producers or wholesalers, the greatest part of the markups are captured by retailers. In relation to this point, Puig-Junoy (2009) 35 finds average discounts on the wholesaler price for generic pravastatins and simvastatins to be above 50% in Spain. In terms of welfare, for the whole market there is a considerable reduction in variable profits. On the other hand, the reduction of prices would increase consumer surplus notably and the net effect would be clearly positive under any of the four scenarios considered. This implies that the liberalization of prices together with pro-competitive regulation would increase welfare. These results are in line with Kanavos, Costa-Font and Seeley (2008) whose evaluation of the eight top pharmaceutical markets worldwide, which included Spain, found that, when generic drugs are on the market, health insurers (such as the NHS in Spain) do not capitalize totally or save costs quickly due to the lower prices of the generic drugs. One of their explanations is that generic drug prices are closely tied to originator brand-name drug prices and that reference pricing reduces generic prices but only marginally. In order to benefit from the introduction of generics, these authors recommend price competition and avoidance of price-fixing regulation that ties generic prices to those of originator brands. Additionally, unrestricted price competition may allow discounts to emerge; which currently seem to be captured mainly by pharmacists. If this were to happen, the saving in costs due to price rebates could benefit the NHS (or tax-payers) and patients. With price competition it seems that the main losers would not be the producers and wholesalers in general but the retailers. 36 Among manufacturers, those producing mainly generics and active ingredients that experience a large reduction in sales, especially simvastatin, would be net losers. In contrast, those selling mainly pravastatin would benefit. Moreover, among firms that produce the same active ingredient, market shares would be distributed more unequally than in the real situation. Similarly to the differences between active ingredients, once the price gap is reduced, other characteristics of the manufacturers should explain those changes. It is important to note that, in the simulated unrestricted price competition scenarios, prices are reduced because firms are assumed to compete à la Bertrand and not to cooperate. In other words, it is assumed that firms do not collude when prices may be freely set. Although collusion is prohibited by Article 101 of the Treaty on the Functioning of the European Union and by the Spanish Law for the Defence of Competition (Act 15/2007), the oligopolistic structure of this industry may facilitate it. Furthermore, it could be argued that cooperation among producers may be relatively high in the pharmaceutical market where a great number of firms meet in different markets. In some sense, the comparison between the real prices and those obtained in the scenarios of perfect collusion between all the producers suggests that the regulation of prices may have the same effect as cooperation between firms. For instance, the reference pricing system tends to cluster prices at the reference price (Danzon and Liu, 1998), which is not always close to the competitive level. This problem may be 37 overcome, at least partially, if the reference price is set as the lowest price in the substitution group, as it is in Denmark (Kaiser, Mendez and Rønde, 2010). However, the relatively high prices of the real situation may be explained by the high prices of entry that are usual in regulated pharmaceutical markets. In the simulation it is assumed that firms always play a one-shot game and therefore they can modify the prices of their products every month: in each period firms set prices taking into account both the fact that they are multi-product firms and the decisions of their competitors in each market, in order to maximize profits. Nevertheless, in the real situation, producers negotiate introductory prices with health authorities and these prices usually remain constant for a long period and decline over time in real terms or are reduced with the implementation of cost containment measures. Thus, given that price increases are uncommon, the NHS compensates the producers with a relatively high introductory price as Ekelund and Persson (2003) found for the Swedish market. In other words, the regulation of prices impedes penetration strategies, i.e., entry into the market with low prices. For that reason, firms usually opt for a skimming strategy, which involves setting a relatively high introductory price. This may explain why pravastatin experienced such a great increase in market share with price competition. It is a relatively new product with the highest average price and so producers have more of a margin over which to reduce their prices when there is unrestricted competition and therefore sales increase. In fact, fluvastatin, another relatively new product, has the second highest 38 average price and is the only active ingredient, except for pravastatin, that does not lose part of its market share. Even with the entry of generic medicines for the first time into the statin market in November 2000, there is currently only limited price competition. One possible explanation is that, as stated by Puig-Junoy (2007), who also analyses the statin market in Spain, the decline in the price of brand-name and generic off-patent products is associated with arbitrary regulatory decisions as to the period for which the product is included in the reference pricing system or the moment at which its reference price is revised. Similarly, Puig-Junoy and Moreno-Torres (2010) concluded that the Spanish reference pricing system results in very little price competition between generic firms and that price reductions are mainly limited to specific regulatory measures. Moreover, it is possible that companies selling to several countries are reluctant to decrease prices in Spain in order to keep high prices in other countries. A low price in Spain may reduce the bargaining position of firms when they negotiate prices or may even lead to a direct reduction in prices in other countries if they are based on foreign prices. For example, the external reference pricing, which imposes a price cap based on prices of identical or comparable products in other countries, is common in European countries. What ever the case, the regulation of prices and the application of cost containment measures have not been totally successful in reducing prices and maximizing 39 welfare. In comparison to the scenarios simulated, it seems that regulation undermines price competition, as was found by Danzon and Chao (2000). 8. Concluding remarks This paper analyses the demand for statin drugs and reports a counterfactual exercise in unrestricted price competition in the regulated Spanish pharmaceutical market. This comparison between the real and the counterfactual scenarios allows me to evaluate the elimination of price rigidity in terms of welfare, and of winners and losers. The simulations of price competition à la Bertrand under four different scenarios indicate that in a situation with freedom to set prices and a pro-competitive environment of prices, there would be a general reduction in prices. The price reduction for the whole market would range from 39% to 59%, depending on the assumptions regarding the products that compete and the marginal cost. Due to the negative price elasticity of the demand, there would be an increase in the total quantity of DDDs consumed. However, the decline in prices would more than compensate for it and there would be a considerable expenditure decrease. The savings for the NHS and patients would range from 37% to 57%. As a consequence of price variations, there would be important changes in the total number of DDDs and market shares among active ingredients. Also, the 40 counterfactual exercise shows a more unequal distribution of sales among producers and a general reversal of generic drugs. Finally, there would be a global reduction in revenue and, in fact, most firms would experience reductions in variable profits. For the market as a whole, the reduction in variable profits would range from 36% to 57%. However, the reduction in prices would increase consumer surplus and the net effect would be clearly positive under all the scenarios. Provided that there is no collusion between firms, the liberalization of prices would enhance welfare and may cause rebates (which at present seem to be captured mainly by retailers) to be reflected in final prices and thus transfer those benefits to the NHS (tax-payers) and patients in the form of cost savings. However, it should be considered that this would involve not only winners but also losers. In conclusion, in comparison to the scenarios simulated, it seems that regulation of prices, such as reference pricing, may have a similar effect to cooperation among producers and undermine price competition. More freedom to set prices is necessary together with a regulatory framework that includes appropriate incentives to foster competition. 41 Acknowledgments The author would like to thank Jaume Puig-Junoy and Joan-Ramon Borrell-Arqué for their guidance and support and Panos Kanavos, Joan Costa-Font and Javier Coronado for helpful comments. I am grateful to the Directorate-General of Pharmacy and Health Products of the Spanish Ministry of Health and Consumer Affairs for data used in this paper and to the LSE Health Centre, London School of Economics and Political Science, London, UK, where part of this paper was developed during a stay in the summer of 2008, which was supported by a BE-DGR grant from the University and Research Grant Management Agency (AGAUR) of the Autonomous Government of Catalonia. I also benefited from the support of grant SEJ2007-66133 from the Spanish Ministry of Education and Science and from the support of an unrestricted educational grant from the Merck Company Foundation, the philanthropic arm of Merck & Co. Inc., Whitehouse Station, New Jersey, USA. The opinions expressed are those of the author and do not necessarily reflect the opinions of the research sponsors. 42 References [1] Antoñanzas F, Oliva J, Pinillos M, Juárez C. Economic aspects of the new Spanish laws on pharmaceutical preparations. European Journal Health Economics 2007; 8: 297–300. Aronsson T, Bergman M, Rudholm N. The impact of generic drug competition on brand-name market shares–Evidence from micro data. Review of Industrial Organization 2001; 19: 425–435. Baye MR, Kovenock D. Bertrand competition. The New Palgrave Dictionary of Economics. Second Edition. Eds. Steven N. Durlauf and Lawrence E. Blume. Palgrave Macmillan, 2008. Berry ST. Estimating Discrete-Choice Models of Product Differentiation. RAND Journal of Economics 1994; 25(2): 242–262. Berry ST, Levinsohn J, Pakes A. Automobile Prices in Market Equilibrium. Econometrica 1995; 63: 841-890. Besanko D, Gupta S, Jain D. Logit Demand Estimation under Competitive Pricing Behavior: An Equilibrium Framework. Management Science 1998; 44(11, Part 1 of 2): 1533–1547. Borrell JR. Drug price differentials caused by formularies and price caps. International Journal of the Economics of Business 2003; 10: 37–50. Borrell JR, Merino-Castelló A. Los beneficios de una competencia incipiente: descuentos y bonificaciones a oficinas de farmacia, en Lluís Cases (Ed.) Anuario de la Competencia 2006, Fundación ICO-Marcial Pons, Madrid, 2007. Bresnahan T, Stern S, Trajtenberg M. Market Segmentation and the Sources of Rents from Innovation: Personal Computers in the Late 1980’s. RAND Journal of Economics 1997; 28: S17–S44. Cardell NS. Variance Components Structures for the Extreme-Value and Logistic Distributions with Application to Models of Heterogeneity. Econometric Theory 1997; 13: 185–213. Cleanthous P. Evaluating innovation in the pharmaceutical industry. Mimeo, New York Univeristy, 2004. Collins R, Armitage J, Parish S, Sleigh P, Peto R. Heart Protection Study Collaborative Group. MRC/BHF Heart Protection Study of cholesterollowering with simvastatin in 5963 people with diabetes: a randomised placebo-controlled trial. Lancet 2003; 361: 2005–2016. [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] 43 [13] Consejo General de Colegios Oficiales de Farmacéuticos. BOT PLUS: Base de datos del conocimiento sanitario. Consejo General de Colegios Oficiales de Farmacéuticos: Madrid, 2005. Coronado, FJ. Market Structure and Regulation in Pharmaceutical Markets. Doctoral thesis dissertation. Universitat Pompeu Fabra: Barcelona, 2010. Coscelli A. The importance of Doctors’ and Patients’ Preferences in the Prescription Decision. Journal of Industrial Economics 2000; 3: 349–369. Costa-Font J, Puig-Junoy J. The pharmaceutical market regulation in Spain: is drug cost-containment under question? Journal of Pharmaceuticals Finance, Economics and Policy 2005; 13: 33–49. Dalen DM, Strøm S, Haabeth T. Price regulation and generic competition in the pharmaceutical market. European Journal of Health Economics 2006; 7: 155–220. Danzon, PM, Chao LW. Does Regulation Drive out Competition in Pharmaceutical Markets? Journal of Law and Economics 2000; 43(2): 311– 358. Danzon PM, Liu H. Reference pricing and physician drug budgets: the German experience in controlling pharmaceutical expenditures. Working paper, The Wharton School, University of Pennsylvania, 1998. Davis P. Spatial competition in retail markets: movie theatres. RAND Journal of Economics 2006; 37(4): 964–982. Di Giacomo M. GMM estimation of a structural demand model for yogurt and the effects of the introduction of new brands. Empirical Economics 2008; 34: 537–565. Ekelund M, Persson B. Pharmaceutical pricing in a regulated market. Review of Economics and Statistics 2003; 85(2): 298–306. Ellison S, Cockburn I, Griliches Z, Hausman J. Characteristics of Demand for Pharmaceutical Products: An Examination of Four Cephalosporins. RAND Journal of Economics 1997; 28: 426–446. Frank RG, Salkever DS. Generic entry and the pricing of pharmaceuticals. Journal of Economics and Management Strategy 1997; 6: 75– 90. Grabowski HG, Vernon JM. Brand loyalty, entry, and price competition in pharmaceuticals after the 1984 drug act. Journal of Law and Economics 1992; 35: 331–350. [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] 44 [26] Hanai J, Cao P, Tanksale P, Imamura S, Koshimizu E, Zhao J, Kishi S, Yamashita M, Phillips PS, Sukhatme VP, Lecker SH. The muscle-specific ubiquitin ligase atrogin-1/MAFbx mediates statin-induced muscle toxicity. Journal of Clinical Investigation 2007; 117(12): 3940–3951. Hansen L. Large sample properties of generalized method of moments estimators. Econometrica 1982; 50(3): 1029–1054. Hausman JA, Leonard GK. The competitive effects of a new product introduction: a case study. Journal of Industrial Economics 2002; L3: 237–263. Hellerstein JK. The importance of the physician in the generic versus trade-name prescription decision. RAND Journal of Economics 1998; 29(1): 108–136. Iizuka T. Experts’ agency problems: evidence from the prescription drug market in Japan. RAND Journal of Economics 2007; 38(3): 844–862. IMS Health. Estudio del impacto de los medicamentos genéricos en el mercado español. IMS Health: Madrid, 2006. Kaiser U, Mendez SJ, Rønde T. Regulation of pharmaceutical prices: evidence from a reference price reform in Denmark. ZEW Discussion Paper No. 10-062, 2010. (ftp://ftp.zew.de/pub/zewdocs/dp/dp10062.pdf). Kanavos P, Costa-Font J, Seeley E. Competition in off-patent drug markets: Issues, regulation and evidence. Economic Policy 2008; 23(55): 499–544. Kanavos P, Taylor D. Pharmacy discounts on generic medicines in France: is there room for further efficiency savings? Current Medical Research and Opinion 2007; 23(10): 2467–2476. López-Casasnovas G, Puig-Junoy J. Review of the literature on reference pricing. Health Policy 2000; 54: 87–123. Ministerio de Sanidad y Consumo. Nomenclator Digitalis. Instituto de Información Sanitaria del Sistema Nacional de Salud. Ministerio de Sanidad y Consumo: Madrid, 2005. Moreno-Torres I, Puig-Junoy J, Borrell JR. Generic Entry into the Regulated Spanish Pharmaceutical Market. Review of Industrial Organization 2009; 34(4): 373–388. Nevo A. Mergers with Differentiated Products: The Case of the Readyto-Eat Cereal Industry. RAND Journal of Economics 2000a; 31: 395–421. [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] 45 [39] Nevo A. A practitioner's guide to estimation of random-coeffients logit models of demand. Journal of Economics & Management Strategy 2000b; 9(4): 513–548. Nevo A. Measuring Market Power in the Ready-to-Eat Cereal Industry. Econometrica 2001; 69: 307–342. Petrin A. Quantifying the benefits of new products: the case of the Minivan. Journal of Political Economy 2002; 110: 705–729. Puig-Junoy J. The impact of generic reference pricing interventions in the statin market. Health Policy 2007; 84(1): 14–29. Puig-Junoy J. Impacto de la regulación de precio de los medicamentos sobre la competencia en el mercado de genéricos: valoración de los efectos y necesidad de reforma. Autoritat Catalana de la Competència: Barcelona, 2009. Puig-Junoy J, Moreno-Torres I. Do generic firms and the Spanish public purchaser respond to consumer price differences of generics under reference pricing? Health Policy 2010; 98(2): 186–194. Reiffen D, Ward ME. Generic drug industry dynamics. Review of Economics and Statistics 2005; 87: 37–49. Sargan J. The estimation of economic relationships using instrumental variables. Econometrica 1958; 26(3): 393–415. Scherer FM. Pricing, profits, and technological progress in the pharmaceutical industry. Journal of Economic Perspectives 1993; 7(3): 97–115. Shepherd J, Hunninghake DB, Barter P, McKenney JM, Hutchinson HG. Guidelines for lowering lipids to reduce coronary artery disease risk: a comparison of rosuvastatin with atorvastatin, pravastatin, and simvastatin for achieving lipid-lowering goals. American Journal of Cardiology 2003; 91(5A): 11C–17C; discussion 17C–19C. Small KA, Rosen HS. Applied Welfare Economics with Discrete Choice Models. Econometrica 1981; 49(1): 105–130. Stern S. Market Definition and the Returns to Innovation: Substitution Patterns in Pharmaceutical Markets. MIT POPI Working Paper. 1-51. 1996. Verboven F. International Price Discrimination in the European Car Market. RAND Journal of Economics 1996; 27: 240–268. [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] 46 [52] Yu Y, Gupta S. Pioneering Advantage in Generic Drug Competition. Cornell University. Johnson School Research Paper Series No. 37-06. October 2008. 47 Figures and Tables Figure 1. Drug market Statins Outside good Simvastatin Lovastatin Pravastatin Fluvastatin Atorvastatin Cerivastatin Figure 2. Active ingredients: total amount (€) for 2005 Total sales amount 140000000 120000000 100000000 80000000 60000000 40000000 20000000 0 Real All (mc=.07) All (mc=0) Off-patent (mc=.07) Off-patent (mc=0) ATORVASTATIN FLUVASTATIN LOVASTATIN PRAVASTATIN SIMVASTATIN TOTAL 48 Figure 3. Active ingredient market shares (€) for 2005 Active ingredients market shares (amount) 0,60 0,50 0,40 0,30 0,20 0,10 0,00 Real All (mc=.07) All (mc=0) Off-patent (mc=.07) Off-patent (mc=0) ATORVASTATIN FLUVASTATIN LOVASTATIN PRAVASTATIN SIMVASTATIN Figure 4. Active ingredients: total quantity (DDDs) for 2005 Active ingredients quantities (DDDs) 300000000 250000000 ATORVASTATIN 200000000 150000000 100000000 50000000 0 Real All (mc=.07) All (mc=0) Off-patent (mc=.07) Off-patent (mc=0) FLUVASTATIN LOVASTATIN PRAVASTATIN SIMVASTATIN TOTAL 49 Figure 5. Active ingredient market shares (DDDs) for 2005 Active ingredients market share (quantity) 1,00 0,90 0,80 0,70 ATORVASTATIN 0,60 0,50 0,40 0,30 0,20 0,10 0,00 Real All (mc=.07) All (mc=0) Off-patent (mc=.07) Off-patent (mc=0) FLUVASTATIN LOVASTATIN PRAVASTATIN SIMVASTATIN Figure 6. Brand-name and generic drug market shares (€) for 2005 Total market shares (amount) 1,00 0,90 0,80 0,70 0,60 0,50 0,40 0,30 0,20 0,10 0,00 Real All (mc=.07) All (mc=0) Off-patent (mc=.07) Off-patent (mc=0) Brand Gen. 50 Figure 7. Brand-name and generic drug market shares (DDDs) for 2005 Total market shares (quantity) 1,00 0,90 0,80 0,70 0,60 0,50 0,40 0,30 0,20 0,10 0,00 Real All (mc=.07) All (mc=0) Off-patent (mc=.07) Off-patent (mc=0) Brand Gen. Figure 8. Market shares of atorvastatin manufacturers (DDDs) in 2005 Atorvastatin firms market shares 0,36 0,35 0,34 0,33 0,32 0,31 0,30 0,29 Real All (mc=.07) All (mc=0) Off-patent (mc=.07) Off-patent (mc=0) 51 Figure 9. Market shares of fluvastatin manufacturers (DDDs) in 2005 Fluvastatin firms market shares 0,35 0,30 0,25 0,20 0,15 0,10 0,05 0,00 Real All (mc=.07) All (mc=0) Off-patent (mc=.07) Off-patent (mc=0) Figure 10. Market shares of lovastatin manufacturers (DDDs) in 2005 Lovastatin firms market shares 0,45 0,40 0,35 0,30 0,25 0,20 0,15 0,10 0,05 0,00 Real All (mc=.07) All (mc=0) Off-patent (mc=.07) Off-patent (mc=0) 52 Figure 11. Market shares of pravastatin manufacturers (DDDs) in 2005 Pravastatin firms market shares 0,60 0,50 0,40 0,30 0,20 0,10 0,00 Real All (mc=.07) All (mc=0) Off-patent (mc=.07) Off-patent (mc=0) Figure 12. Market shares of simvastatin producer manufacturers (DDDs) in 2005 Simvastatin firms market shares 0,45 0,40 0,35 0,30 0,25 0,20 0,15 0,10 0,05 0,00 Real All (mc=.07) All (mc=0) Off-patent (mc=.07) Off-patent (mc=0) 53 Figure 13. Market shares of atorvastatin manufacturers (€) in 2005 Atorvastatin firms (value) market shares 0,35 0,34 0,34 0,33 0,33 0,32 0,32 Real All (mc=.07) All (mc=0) Off-patent (mc=.07) Off-patent (mc=0) Figure 14. Market shares of fluvastatin manufacturers (€) in 2005 Fluvastatin firms (value) market shares 0,35 0,30 0,25 0,20 0,15 0,10 0,05 0,00 Real All (mc=.07) All (mc=0) Off-patent (mc=.07) Off-patent (mc=0) 54 Figure 15. Market shares of lovastatin manufacturers (€) in 2005 Lovastatin firms (value) market shares 0,45 0,40 0,35 0,30 0,25 0,20 0,15 0,10 0,05 0,00 Real All (mc=.07) All (mc=0) Off-patent (mc=.07) Off-patent (mc=0) Figure 16. Market shares of pravastatin manufacturers (€) in 2005 Pravastatin firms (value) market shares 0,60 0,50 0,40 0,30 0,20 0,10 0,00 Real All (mc=.07) All (mc=0) Off-patent (mc=.07) Off-patent (mc=0) 55 Figure 17. Market shares of simvastatin manufacturers (€) in 2005 Simvastatin firms (value) market shares 0,50 0,45 0,40 0,35 0,30 0,25 0,20 0,15 0,10 0,05 0,00 Real All (mc=.07) All (mc=0) Off-patent (mc=.07) Off-patent (mc=0) 56 Table 1. Sample of drugs Active Ingredient (ATC code) Main indications First Obs. November 1997 May 2001 May 2001 January 2003 August 1998 August 1998 October 2002 November 2000 January 1997 January 1997 September 2002 January 1997 January 1997 January 1997 January 1997 February 1999 January 1997 January 1997 September 1997 Generic entry November 2000 December 2000 January 2004 January 2004 January 2004 January 2002 January 2002 January 2002 Reference pricing May 2002 May 2002 January 2004 January 2004 January 2004 Presentations 10 mg 20 mg 40 mg 80 mg 0.1 mg 0.2 mg 0.3 mg 0.4 mg 20 mg 40 mg 80 mg 20 mg 40 mg 10 mg 20 mg 40 mg 10 mg 20 mg 40 mg 28 Tablets 28 Tablets 28 Tablets 28 Tablets 28 Tablets 28 Tablets 28 Tablets 28 Tablets 28 Capsules 28 Capsules 28 Capsules 28 Tablets 28 Tablets 28 Tablets 28 Tablets 28 Tablets 28 Tablets 28 Tablets 28 Tablets Atorvastatin (C10AA05) Dyslipidemia, hipercolesterolemia and hypertriglyceridemia. Cerivastatin (C10AA06) Hipercolesterolemia. Fluvastatin (C10AA04) Dyslipidemia, hypercholesterolemia, hypertriglyceridemia and atherosclerotic. Dyslipidemia, hypercholesterolemia, hyperlipoproteinemia and atherosclerotic. Dyslipidemia, hypercholesterolemia, hypertriglyceridemia, Atherosclerotic, Ischemic cardiomyopathy and acute myocardial infarction. Dyslipidemia, hypercholesterolemia, hypertriglyceridemia, Atherosclerotic, Ischemic cardiomyopathy and diabetes. Lovastatin (C10AA02) Pravastatin (C10AA03) Simvastatin (C10AA01) 57 Table 2. Summary of statistics for the variables in the demand model. Variable Product market share Outside good market share Price per DDD Within active ingredient share Definition Product share in the potential market Outside good share in the potential market Price per defined daily dose (€/DDD) Generic drug’s market share divided by the total generic market share Number of months since the entry of the generic drug Dummy equal to one for generics; 0 otherwise Dummy equal to one for cerivastatins from the withdrawal; 0 otherwise Dummy equal to one for atorvastatins; 0 otherwise Dummy equal to one for cerivastatins; 0 otherwise Dummy equal to one for fluvastatins; 0 otherwise Dummy equal to one for lovastatins; 0 otherwise Dummy equal to one for pravastatins; 0 otherwise Obs. 10981 10981 10981 Mean 0.0038 0.4546 0.7880 Standard Deviation 0.0072 0.2004 0.3716 Minimum 0.0000 0.1833 0.1882 Maximum 0.0544 0.9135 1858022 10981 0.0533 0.0908 0.0000 0.8837 Time in market Generic Cerivastatin withdrawal 10981 10981 5257108 0.4586 387245 0.4983 0 0 181 1 10981 0.0101 0.1000 0 1 Atorvastatin Cerivastatin Fluvastatin Lovastatin Pravastatin 10981 10981 0.0670 0.0387 0.2501 0.1929 0 0 1 1 10981 10981 0.0849 0.2325 0.2787 0.4224 0 0 1 1 10981 0.1616 0.3681 0 1 58 Table 3. Demand model estimation. Coefficient Log(Within active ingredient share) Log(Price per DDD) Time in market Generic Cerivastatin withdrawal Atorvastatin Cerivastatin Fluvastatin Lovastatin Pravastatin Constant Observations R2 Sargan-Hansen statistic (p-value) 0.4580*** -3.4569*** 0.0105*** 0.5621** -7.5984*** 4.8069*** 4.1433*** 3.0647*** -1.0277*** 0.9457*** -4.6815*** 10981 0.8392 0.0680 (0.7950) Std. Error♦ 0.1049 0.2431 0.0017 0.2611 0.2670 1.0194 1.1738 0.7986 0.1404 0.1793 1.2116 Note: Fixed firm effects are not shown but are included in the estimation. ♦ Robust clustered standard errors. ** and *** = significant at the 5% and 1% level. Table 4. Active ingredients: weighted average prices (€/DDD) for 2005 Real MC = 0.07 Atorvastatin Fluvastatin Lovastatin Pravastatin Simvastatin Statins 0.36 0.77 0.55 1.00 0.37 0.44 0.26 0.24 0.24 0.26 0.25 0.25 All MC = 0 0.19 0.17 0.17 0.19 0.18 0.18 Off-patent MC = 0.07 MC = 0 0.36 0.77 0.24 0.26 0.25 0.27 0.36 0.77 0.17 0.19 0.18 0.21 Table 5. Active ingredients: total amount (€) for 2005 Real Atorvastatin Fluvastatin Lovastatin Pravastatin Simvastatin Statins Total variation (∆%) ∆ 36324392 5603596 9572070 22710510 46990844 121201412.00 MC = 0.07 18755274 3809753 3120487 34466320 12194016 72345850.00 -48855562.00 (-40.31%) All MC = 0 13691786 2720561 2224253.25 25161054 8794725 52592379.25 -68609032.75 (-56.61%) Off-patent MC = 0.07 MC = 0 11637822 9415679 1795859.125 1452968.125 3971503 2909934.25 43889544 32988476 15525136 11498549 76819864.13 58265606.38 -44381547.88 -62935805.63 (-36.62%) (-51.93%) 59 Table 6. Active ingredient market shares (€) for 2005 Real MC = 0.07 Atorvastatin Fluvastatin Lovastatin Pravastatin Simvastatin 0.30 0.05 0.08 0.19 0.39 0.26 0.05 0.04 0.48 0.17 All MC = 0 0.26 0.05 0.04 0.48 0.17 Off-patent MC = 0.07 MC = 0 0.15 0.02 0.05 0.57 0.20 0.16 0.02 0.05 0.57 0.20 Table 7. Active ingredients: total quantity (DDDs) for 2005 Real MC = 0.07 Atorvastatin Fluvastatin Lovastatin Pravastatin Simvastatin Statins Total variation (∆%) ∆ 100779872 7247525 17378418 22783328 125385152 273574295 72863240 15664540 12888991 133904352 48900064 284221187 10646892.00 (3.89%) All MC = 0 73058784 15706754 12923670 134261536 49031480 284982224 11407929.00 (4.17%) Off-patent MC = 0.07 MC = 0 32292750 2323378.50 16428957 169790880 62408512 283244477.50 9670182.50 (3.53%) 26126828 1879782 16910576 174895072 64301640 284113898.00 10539603 (3.85%) Table 8. Active ingredient market shares (DDDs) for 2005 Real Atorvastatin Fluvastatin Lovastatin Pravastatin Simvastatin 0.37 0.03 0.06 0.08 0.46 MC = 0.07 0.26 0.06 0.05 0.47 0.17 All MC = 0 0.26 0.06 0.05 0.47 0.17 Off-patent MC = 0.07 MC = 0 0.11 0.09 0.01 0.01 0.06 0.06 0.60 0.62 0.22 0.23 Table 9. Brand-name and generic drug market shares (€) for 2005 Real Lovastatin Pravastatin Simvastatin Statins Brand-name Generic Brand-name Generic Brand-name Generic Brand-name Generic 0.89 0.11 0.66 0.34 0.86 0.14 0.87 0.13 All MC = 0.07 0.90 0.10 0.94 0.06 0.88 0.12 0.95 0.05 MC = 0 0.90 0.10 0.95 0.05 0.88 0.12 0.95 0.05 Off-patent MC = 0.07 MC = 0 0.90 0.90 0.10 0.10 0.94 0.95 0.06 0.05 0.88 0.88 0.12 0.12 0.94 0.94 0.06 0.06 60 Table 10. Brand-name and generic drug market shares (DDDs) for 2005 Real Lovastatin Pravastatin Simvastatin Statins Brand-name Generic Brand-name Generic Brand-name Generic Brand-name Generic 0.88 0.12 0.59 0.41 0.84 0.16 0.88 0.12 All MC = 0.07 0.89 0.11 0.94 0.06 0.87 0.13 0.94 0.06 MC = 0 0.89 0.11 0.94 0.06 0.87 0.13 0.94 0.06 Off-patent MC = 0.07 MC = 0 0.89 0.89 0.11 0.11 0.94 0.93 0.06 0.07 0.87 0.87 0.13 0.13 0.93 0.92 0.07 0.08 Table 11. Variations in variable profits (€) in 2005 All Atorvastatin Fluvastatin Lovastatin Pravastatin Simvastatin Total variation (∆%) ∆ MC = 0.07 -16840717.84 -1793843.03 -6451582.76 11755809.94 -34796828.15 -48127161.83 (-40.49%) MC = 0 -22632608.00 -2883035.17 -7347816.50 2450544.29 -38196118.95 -68609034.33 (-56.61%) Off-patent MC = 0.07 MC = 0 -23095765.05 -26908714.75 -3807736.97 -4150627.93 -5600566.89 -6662135.55 21179033.42 10277965.55 -31465707.95 -35492295.07 -42790743.44 -62935807.75 (-36.00%) (-51.93%) Table 12. Variation in welfare (€) in 2005 All Consumers surplus Firms’ profits Net variation MC = 0.07 669688171.48 -16290016.45 653398155.03 MC = 0 785952603.11 -68609034.33 717343568.78 Off-patent MC = 0.07 MC = 0 552526000.62 655479835.08 -14043786.56 -62935807.75 538482214.06 592544027.33 Table 13. Active ingredients: weighted average full collusion prices (€/DDD) for 2005 Real MC = 0.07 Atorvastatin Fluvastatin Lovastatin Pravastatin Simvastatin Statins 0.36 0.77 0.55 1.00 0.37 0.44 0.48 0.48 0.48 0.48 0.48 0.48 All MC = 0 0.41 0.41 0.41 0.41 0.41 0.41 Off-patent MC = 0.07 MC = 0 0.36 0.77 0.45 0.45 0.45 0.45 0.36 0.77 0.38 0.38 0.38 0.38 61 SÈRIE DE DOCUMENTS DE TREBALL DE LA XREAP 2006 CREAP2006-01 Matas, A. (GEAP); Raymond, J.Ll. (GEAP) "Economic development and changes in car ownership patterns" (Juny 2006) CREAP2006-02 Trillas, F. (IEB); Montolio, D. (IEB); Duch, N. (IEB) "Productive efficiency and regulatory reform: The case of Vehicle Inspection Services" (Setembre 2006) CREAP2006-03 Bel, G. (PPRE-IREA); Fageda, X. (PPRE-IREA) "Factors explaining local privatization: A meta-regression analysis" (Octubre 2006) CREAP2006-04 Fernàndez-Villadangos, L. (PPRE-IREA) "Are two-part tariffs efficient when consumers plan ahead?: An empirical study" (Octubre 2006) CREAP2006-05 Artís, M. (AQR-IREA); Ramos, R. (AQR-IREA); Suriñach, J. (AQR-IREA) "Job losses, outsourcing and relocation: Empirical evidence using microdata" (Octubre 2006) CREAP2006-06 Alcañiz, M. (RISC-IREA); Costa, A.; Guillén, M. (RISC-IREA); Luna, C.; Rovira, C. "Calculation of the variance in surveys of the economic climate” (Novembre 2006) CREAP2006-07 Albalate, D. (PPRE-IREA) "Lowering blood alcohol content levels to save lives: The European Experience” (Desembre 2006) CREAP2006-08 Garrido, A. (IEB); Arqué, P. (IEB) “The choice of banking firm: Are the interest rate a significant criteria?” (Desembre 2006) SÈRIE DE DOCUMENTS DE TREBALL DE LA XREAP CREAP2006-09 Segarra, A. (GRIT); Teruel-Carrizosa, M. (GRIT) "Productivity growth and competition in spanish manufacturing firms: What has happened in recent years?” (Desembre 2006) CREAP2006-10 Andonova, V.; Díaz-Serrano, Luis. (CREB) "Political institutions and the development of telecommunications” (Desembre 2006) CREAP2006-11 Raymond, J.L.(GEAP); Roig, J.L.. (GEAP) "Capital humano: un análisis comparativo Catalunya-España” (Desembre 2006) CREAP2006-12 Rodríguez, M.(CREB); Stoyanova, A. (CREB) "Changes in the demand for private medical insurance following a shift in tax incentives” (Desembre 2006) CREAP2006-13 Royuela, V. (AQR-IREA); Lambiri, D.; Biagi, B. "Economía urbana y calidad de vida. Una revisión del estado del conocimiento en España” (Desembre 2006) CREAP2006-14 Camarero, M.; Carrion-i-Silvestre, J.LL. (AQR-IREA).;Tamarit, C. "New evidence of the real interest rate parity for OECD countries using panel unit root tests with breaks” (Desembre 2006) CREAP2006-15 Karanassou, M.; Sala, H. (GEAP).;Snower , D. J. "The macroeconomics of the labor market: Three fundamental views” (Desembre 2006) SÈRIE DE DOCUMENTS DE TREBALL DE LA XREAP 2007 XREAP2007-01 Castany, L (AQR-IREA); López-Bazo, E. (AQR-IREA).;Moreno , R. (AQR-IREA) "Decomposing differences in total factor productivity across firm size” (Març 2007) XREAP2007-02 Raymond, J. Ll. (GEAP); Roig, J. Ll. (GEAP) “Una propuesta de evaluación de las externalidades de capital humano en la empresa" (Abril 2007) XREAP2007-03 Durán, J. M. (IEB); Esteller, A. (IEB) “An empirical analysis of wealth taxation: Equity vs. Tax compliance” (Juny 2007) XREAP2007-04 Matas, A. (GEAP); Raymond, J.Ll. (GEAP) “Cross-section data, disequilibrium situations and estimated coefficients: evidence from car ownership demand” (Juny 2007) XREAP2007-05 Jofre-Montseny, J. (IEB); Solé-Ollé, A. (IEB) “Tax differentials and agglomeration economies in intraregional firm location” (Juny 2007) XREAP2007-06 Álvarez-Albelo, C. (CREB); Hernández-Martín, R. “Explaining high economic growth in small tourism countries with a dynamic general equilibrium model” (Juliol 2007) XREAP2007-07 Duch, N. (IEB); Montolio, D. (IEB); Mediavilla, M. “Evaluating the impact of public subsidies on a firm’s performance: a quasi-experimental approach” (Juliol 2007) XREAP2007-08 Segarra-Blasco, A. (GRIT) “Innovation sources and productivity: a quantile regression analysis” (Octubre 2007) SÈRIE DE DOCUMENTS DE TREBALL DE LA XREAP XREAP2007-09 Albalate, D. (PPRE-IREA) “Shifting death to their Alternatives: The case of Toll Motorways” (Octubre 2007) XREAP2007-10 Segarra-Blasco, A. (GRIT); Garcia-Quevedo, J. (IEB); Teruel-Carrizosa, M. (GRIT) “Barriers to innovation and public policy in catalonia” (Novembre 2007) XREAP2007-11 Bel, G. (PPRE-IREA); Foote, J. “Comparison of recent toll road concession transactions in the United States and France” (Novembre 2007) XREAP2007-12 Segarra-Blasco, A. (GRIT); “Innovation, R&D spillovers and productivity: the role of knowledge-intensive services” (Novembre 2007) XREAP2007-13 Bermúdez Morata, Ll. (RFA-IREA); Guillén Estany, M. (RFA-IREA), Solé Auró, A. (RFA-IREA) “Impacto de la inmigración sobre la esperanza de vida en salud y en discapacidad de la población española” (Novembre 2007) XREAP2007-14 Calaeys, P. (AQR-IREA); Ramos, R. (AQR-IREA), Suriñach, J. (AQR-IREA) “Fiscal sustainability across government tiers” (Desembre 2007) XREAP2007-15 Sánchez Hugalbe, A. (IEB) “Influencia de la inmigración en la elección escolar” (Desembre 2007) SÈRIE DE DOCUMENTS DE TREBALL DE LA XREAP 2008 XREAP2008-01 Durán Weitkamp, C. (GRIT); Martín Bofarull, M. (GRIT) ; Pablo Martí, F. “Economic effects of road accessibility in the Pyrenees: User perspective” (Gener 2008) XREAP2008-02 Díaz-Serrano, L.; Stoyanova, A. P. (CREB) “The Causal Relationship between Individual’s Choice Behavior and Self-Reported Satisfaction: the Case of Residential Mobility in the EU” (Març 2008) XREAP2008-03 Matas, A. (GEAP); Raymond, J. L. (GEAP); Roig, J. L. (GEAP) “Car ownership and access to jobs in Spain” (Abril 2008) XREAP2008-04 Bel, G. (PPRE-IREA) ; Fageda, X. (PPRE-IREA) “Privatization and competition in the delivery of local services: An empirical examination of the dual market hypothesis” (Abril 2008) XREAP2008-05 Matas, A. (GEAP); Raymond, J. L. (GEAP); Roig, J. L. (GEAP) “Job accessibility and employment probability” (Maig 2008) XREAP2008-06 Basher, S. A.; Carrión, J. Ll. (AQR-IREA) Deconstructing Shocks and Persistence in OECD Real Exchange Rates (Juny 2008) XREAP2008-07 Sanromá, E. (IEB); Ramos, R. (AQR-IREA); Simón, H. Portabilidad del capital humano y asimilación de los inmigrantes. Evidencia para España (Juliol 2008) XREAP2008-08 Basher, S. A.; Carrión, J. Ll. (AQR-IREA) Price level convergence, purchasing power parity and multiple structural breaks: An application to US cities (Juliol 2008) XREAP2008-09 Bermúdez, Ll. (RFA-IREA) A priori ratemaking using bivariate poisson regression models (Juliol 2008) SÈRIE DE DOCUMENTS DE TREBALL DE LA XREAP XREAP2008-10 Solé-Ollé, A. (IEB), Hortas Rico, M. (IEB) Does urban sprawl increase the costs of providing local public services? Evidence from Spanish municipalities (Novembre 2008) XREAP2008-11 Teruel-Carrizosa, M. (GRIT), Segarra-Blasco, A. (GRIT) Immigration and Firm Growth: Evidence from Spanish cities (Novembre 2008) XREAP2008-12 Duch-Brown, N. (IEB), García-Quevedo, J. (IEB), Montolio, D. (IEB) Assessing the assignation of public subsidies: Do the experts choose the most efficient R&D projects? (Novembre 2008) XREAP2008-13 Bilotkach, V., Fageda, X. (PPRE-IREA), Flores-Fillol, R. Scheduled service versus personal transportation: the role of distance (Desembre 2008) XREAP2008-14 Albalate, D. (PPRE-IREA), Gel, G. (PPRE-IREA) Tourism and urban transport: Holding demand pressure under supply constraints (Desembre 2008) SÈRIE DE DOCUMENTS DE TREBALL DE LA XREAP 2009 XREAP2009-01 Calonge, S. (CREB); Tejada, O. “A theoretical and practical study on linear reforms of dual taxes” (Febrer 2009) XREAP2009-02 Albalate, D. (PPRE-IREA); Fernández-Villadangos, L. (PPRE-IREA) “Exploring Determinants of Urban Motorcycle Accident Severity: The Case of Barcelona” (Març 2009) XREAP2009-03 Borrell, J. R. (PPRE-IREA); Fernández-Villadangos, L. (PPRE-IREA) “Assessing excess profits from different entry regulations” (Abril 2009) XREAP2009-04 Sanromá, E. (IEB); Ramos, R. (AQR-IREA), Simon, H. “Los salarios de los inmigrantes en el mercado de trabajo español. ¿Importa el origen del capital humano?” (Abril 2009) XREAP2009-05 Jiménez, J. L.; Perdiguero, J. (PPRE-IREA) “(No)competition in the Spanish retailing gasoline market: a variance filter approach” (Maig 2009) XREAP2009-06 Álvarez-Albelo,C. D. (CREB), Manresa, A. (CREB), Pigem-Vigo, M. (CREB) “International trade as the sole engine of growth for an economy” (Juny 2009) XREAP2009-07 Callejón, M. (PPRE-IREA), Ortún V, M. “The Black Box of Business Dynamics” (Setembre 2009) XREAP2009-08 Lucena, A. (CREB) “The antecedents and innovation consequences of organizational search: empirical evidence for Spain” (Octubre 2009) XREAP2009-09 Domènech Campmajó, L. (PPRE-IREA) “Competition between TV Platforms” (Octubre 2009) SÈRIE DE DOCUMENTS DE TREBALL DE LA XREAP XREAP2009-10 Solé-Auró, A. (RFA-IREA),Guillén, M. (RFA-IREA), Crimmins, E. M. “Health care utilization among immigrants and native-born populations in 11 European countries. Results from the Survey of Health, Ageing and Retirement in Europe” (Octubre 2009) XREAP2009-11 Segarra, A. (GRIT), Teruel, M. (GRIT) “Small firms, growth and financial constraints” (Octubre 2009) XREAP2009-12 Matas, A. (GEAP), Raymond, J.Ll. (GEAP), Ruiz, A. (GEAP) “Traffic forecasts under uncertainty and capacity constraints” (Novembre 2009) XREAP2009-13 Sole-Ollé, A. (IEB) “Inter-regional redistribution through infrastructure investment: tactical or programmatic?” (Novembre 2009) XREAP2009-14 Del Barrio-Castro, T., García-Quevedo, J. (IEB) “The determinants of university patenting: Do incentives matter?” (Novembre 2009) XREAP2009-15 Ramos, R. (AQR-IREA), Suriñach, J. (AQR-IREA), Artís, M. (AQR-IREA) “Human capital spillovers, productivity and regional convergence in Spain” (Novembre 2009) XREAP2009-16 Álvarez-Albelo, C. D. (CREB), Hernández-Martín, R. “The commons and anti-commons problems in the tourism economy” (Desembre 2009) SÈRIE DE DOCUMENTS DE TREBALL DE LA XREAP 2010 XREAP2010-01 García-López, M. A. (GEAP) “The Accessibility City. When Transport Infrastructure Matters in Urban Spatial Structure” (Febrer 2010) XREAP2010-02 García-Quevedo, J. (IEB), Mas-Verdú, F. (IEB), Polo-Otero, J. (IEB) “Which firms want PhDs? The effect of the university-industry relationship on the PhD labour market” (Març 2010) XREAP2010-03 Pitt, D., Guillén, M. (RFA-IREA) “An introduction to parametric and non-parametric models for bivariate positive insurance claim severity distributions” (Març 2010) XREAP2010-04 Bermúdez, Ll. (RFA-IREA), Karlis, D. “Modelling dependence in a ratemaking procedure with multivariate Poisson regression models” (Abril 2010) XREAP2010-05 Di Paolo, A. (IEB) “Parental education and family characteristics: educational opportunities across cohorts in Italy and Spain” (Maig 2010) XREAP2010-06 Simón, H. (IEB), Ramos, R. (AQR-IREA), Sanromá, E. (IEB) “Movilidad ocupacional de los inmigrantes en una economía de bajas cualificaciones. El caso de España” (Juny 2010) XREAP2010-07 Di Paolo, A. (GEAP & IEB), Raymond, J. Ll. (GEAP & IEB) “Language knowledge and earnings in Catalonia” (Juliol 2010) XREAP2010-08 Bolancé, C. (RFA-IREA), Alemany, R. (RFA-IREA), Guillén, M. (RFA-IREA) “Prediction of the economic cost of individual long-term care in the Spanish population” (Setembre 2010) XREAP2010-09 Di Paolo, A. (GEAP & IEB) “Knowledge of catalan, public/private sector choice and earnings: Evidence from a double sample selection model” (Setembre 2010) SÈRIE DE DOCUMENTS DE TREBALL DE LA XREAP XREAP2010-10 Coad, A., Segarra, A. (GRIT), Teruel, M. (GRIT) “Like milk or wine: Does firm performance improve with age?” (Setembre 2010) XREAP2010-11 Di Paolo, A. (GEAP & IEB), Raymond, J. Ll. (GEAP & IEB), Calero, J. (IEB) “Exploring educational mobility in Europe” (Octubre 2010) XREAP2010-12 Borrell, A. (GiM-IREA), Fernández-Villadangos, L. (GiM-IREA) “Clustering or scattering: the underlying reason for regulating distance among retail outlets” (Desembre 2010) XREAP2010-13 Di Paolo, A. (GEAP & IEB) “School composition effects in Spain” (Desembre 2010) XREAP2010-14 Fageda, X. (GiM-IREA), Flores-Fillol, R. “Technology, Business Models and Network Structure in the Airline Industry” (Desembre 2010) XREAP2010-15 Albalate, D. (GiM-IREA), Bel, G. (GiM-IREA), Fageda, X. (GiM-IREA) “Is it Redistribution or Centralization? On the Determinants of Government Investment in Infrastructure” (Desembre 2010) XREAP2010-16 Oppedisano, V., Turati, G. “What are the causes of educational inequalities and of their evolution over time in Europe? Evidence from PISA” (Desembre 2010) XREAP2010-17 Canova, L., Vaglio, A. “Why do educated mothers matter? A model of parental help” (Desembre 2010) SÈRIE DE DOCUMENTS DE TREBALL DE LA XREAP 2011 XREAP2011-01 Fageda, X. (GiM-IREA), Perdiguero, J. (GiM-IREA) “An empirical analysis of a merger between a network and low-cost airlines” (Maig 2011) XREAP2011-02 Moreno-Torres, I. (ACCO, CRES & GiM-IREA) “What if there was a stronger pharmaceutical price competition in Spain? When regulation has a similar effect to collusion” (Maig 2011) xreap@pcb.ub.es